The advent of ChatGPT:

Job Made Easy or Job Loss to Data Analysts

ABIOLA TIMOTHY OWOLABI*, OLUWASEYI OLUWADAMILARE OKUNLOLA,

EMMANUEL TAIWO ADEWUYI, JANET IYABO IDOWU,

OLASUNKANMI JAMES OLADAPO

Department of Statistics, Ladoke Akintola,

University of Technology, Ogbomoso,

OYO STATE,

NIGERIA

*Corresponding Author

Abstract: - Artificial Intelligence (AI) has proven valuable in almost every field of endeavour, including

education, sciences, engineering, technology, medical sciences, and numerous other areas of application.

Despite its widespread usefulness, concerns have arisen about AI potentially displacing jobs due to its highly

advanced capabilities, commonly called the "god-man effect." One remarkable AI product is ChatGPT, a

Chatbot developed by an open AI company in the USA that is capable of engaging in conversations that

resemble human interactions. This study explores the strengths and limitations of ChatGPT for data analysis,

with the primary objective of assessing whether ChatGPT poses a threat to the job of data analysts. An

econometric dataset with a sample size of thirty (30), which consists of one dependent variable and three

independent variables, was simulated. The dataset was intentionally generated with issues like

multicollinearity, outliers, and heteroscedasticity. Subsequently, multiple tests were conducted on the datasets

to confirm the presence of these problems. The ChatGPT 3.5 and 4.0 versions were then used to analyse the

data to examine this chatbot's prowess in performing data analysis. ChatGPT 3.5 and 4.0 accurately predicted

the suitable statistical tool for analyzing the simulated datasets. Both versions of ChatGPT emphasized that the

expertise of a professional data analyst would be necessary. While they could offer guidance on data analysis,

they cannot perform the analysis themselves as they are solely AI models. ChatGPT can help with what to do

next when a data analyst gets stuck. However, they should not be recognized as an authority in making

statistical decisions. Therefore, ChatGPT may not replace data analysts but could make their job easier by

serving as a helpful resource to turn to when they encounter challenges.

Key-Words: - Artificial Intelligence, ChatGPT, Data Analysis, Simulation, Data Analysts.

Received: August 14, 2023. Revised: December 11, 2023. Accepted: February 16, 2024. Published: April 4, 2024.

1 Introduction

There is no doubt that the world is currently agog

with the introduction of Artificial Intelligence (AI).

Artificial intelligence has been found useful in

almost every aspect of human life. During the

pandemic, AI technology was used to deliver food

items to people indoors. It has also been found

helpful in many other areas of life like health,

sports, and academics. Though helpful, some

people are worried that their jobs might be at stake.

It has been projected that automation will lead to

400 and 800 million jobs by 2030, [1]. This is

because AI is faster than humans and even more

accurate in most cases. They do exactly what they

are instructed to do without getting tired or

frustrated, even answering programming questions,

[2]. This ability to process a lot of information with

little or no mistakes has given AI an edge in the job

recruitment world. An example of AI technology

currently making waves all over the world is the

ChatGPT, which was released to the public in late

2022, [3].

ChatGPT is a chatbot developed by the Open

AI Company in the USA. The bot is designed to be

able to hold human-like conversations. That is, it

can converse with humans just like a human being.

The ChatGPT was trained based on datasets from

various sources, including books, websites,

research articles, theses, newspapers, and several

other sources. This dataset contains about 175

billion parameters. However, no information is

WSEAS TRANSACTIONS on COMPUTERS

DOI: 10.37394/23205.2024.23.3

Abiola Timothy Owolabi, Oluwaseyi Oluwadamilare Okunlola,

Emmanuel Taiwo Adewuyi,

Janet Iyabo Idowu, Olasunkanmi James Oladapo

E-ISSN: 2224-2872

Volume 23, 2024

available to the public on the quality of data

obtained from those sources, [4].

When asked questions, the ChatGPT responds

by analysing the dataset earlier discussed using the

concept called Natural Language Processing

(NLP), [4], [5]. NLP is a technique of analysis in

which information relevant to the questions being

asked is extracted from a dataset. Utilising this

technique, the ChatGPT may seem to be a question

god who has mastered everything. That may not be

true; it is merely working on the principle with

which it was trained.

Concerning education, ChatGPT could aid

students' learning and overall productivity by

helping them in a variety of ways. It can provide

useful information and resources, which can

improve language skills, effectiveness, and

motivation. However, the use of ChatGPT should

be complementary and not a substitute for human

interaction and students' diligence in learning and

achieving their academic goals. Students should be

encouraged to exercise caution in using ChatGPT

to verify the credibility of the information sources

it generates. While ChatGPT can engage in human-

to-human conversations, it remains a machine

without genuine human feelings despite its capacity

for such interactions, [6]. The ChatGPT is also

limited in the educational space because it can only

produce information based on the existing dataset

on which it was trained. Until it has been trained

with new information, ChatGPT does not have

access to it. Thus, teachers should focus on

building competence in learning technology,

enabling them to be more effective in their

profession, [7].

One of the impediments of ChatGPT to the

academic world is being unable to take

responsibility for the contents it generates. Thus,

when used for research, it can void the integrity of

such research work. The academia is now confused

about whether ChatGPT should be cited as a co-

author. With this limitation, the ChatGPT can also

be helpful but must not be wholly relied upon, [8].

In Statistics, ChatGPT lacks a profound

understanding of statistical methodologies,

including regression analysis, hypothesis testing,

multivariate analysis, and machine learning

algorithms, and it is unable to identify potential

biases, confounding factors, or outliers that might

affect the validity of the results, [9]. Recent

concerns have been about whether ChatGPT would

help or harm nurses and Doctors, [10], [11].

Recently, a ghostwriter and copywriter for a creative

solutions agency and Adobe employees raised

concerns about how the introduction of AI has led to

job loss and shake-ups, [12].

In the face of millions losing their jobs to AI

tools like ChatGPT, this study investigates the

strengths and limitations of ChatGPT for Data

Analysts. This study also examines whether

ChatGPT poses a threat to Data Analysts by

potentially encroaching on their roles and job

opportunities.

2 Methodology

An econometric dataset is simulated for analysis by

ChatGPT using R-programming 4.0. The objective

is to evaluate the AI tool's strengths and limitations

in data analysis in comparison to human analysts.

2.1 Description of the Dataset

The dataset that was simulated comprises several

linear regression datasets deliberately designed to

demonstrate various issues or violations of classical

linear regression. The regression model is

generated as follows:

     (1)

where 󰇛 󰇜 and  

 

M is a definite matrix, which is an identity matrix.

In the simulated datasets, multicollinearity, outliers,

and heteroscedasticity problems were intentionally

introduced.

2.2 Introducing Multicollinearity,

Outlier, and Heteroscedasticity

In introducing multicollinearity, the explanatory

variables were generated following the works of

[13], [14], [15], [16].

     

         

(2)

Outliers can be introduced in two ways: in the x

direction or the y direction. In this study, an

independent variable is polluted with outliers using

the expression given in [17] as:

jj xxfx 2max,202 

(3)

Where j



[1,2,...,30] is the sampled index whose

size is determined by the percentage of the outlier

of interest.

Since the Heteroscedasticity has to do with the

error terms, it was generated to change with one of

WSEAS TRANSACTIONS on COMPUTERS

DOI: 10.37394/23205.2024.23.3

Abiola Timothy Owolabi, Oluwaseyi Oluwadamilare Okunlola,

Emmanuel Taiwo Adewuyi,

Janet Iyabo Idowu, Olasunkanmi James Oladapo

E-ISSN: 2224-2872

Volume 23, 2024

the explanatory variables,

. This is a major

violation of Homoscedasticity. The error term is

generated with the following steps:

Generate standard normal, z. Set ei = zi × σ2. Since

the mean of the error term is 0, set

i



. To

invoke heteroskedasticity so that the variance of the

error term is not constant over

The heteroscedastic error becomes

3ijj xze 

(4)

The Error ei from (4) is modified to follow

AR(1) as with an autocorrelation value ρ:

given

 

iiii euuuniu  1

11 1.,,2,1;





(5)

The ui is added to the



vector to generate

the y, the dependent variable.

2.3 Test for Multiple Linear Regression

Assumptions

Various tests were conducted to confirm the

presence of multicollinearity, outliers, and

heteroscedasticity problems introduced earlier in

the datasets and to verify the violation of

assumptions of the multiple linear regression

(MLR) model.

The test for multicollinearity using Variance

Inflation Factor (VIF) shows x1=2094.253609, x2

= 1.040304, and x3 = 2092.433043. A VIF value

greater than 10 indicates the presence of severe

multicollinearity. The test for Autocorrelation

using the Durbin-Watson test shows the Durbin-

Watson value (DW) = 0.42608 and p-value =

8.226e-09. This shows the presence of

autocorrelation. Studentized Breusch-Pagan test for

Heteroscedasticity gives the value 2.9299 and p-

value = 0.4026. This confirms the presence of

Heteroscedasticity. Cook’s distance, [18] and

dffits diagnostic tools, [19], for detecting outliers

show some observations as outliers in the datasets.

The detected observations are bolded in Table 1

(Appendix).

3 Data Analysis

For analysis purposes, two versions of ChatGPT

(3.5 and 4.0) were utilized to determine whether

there would be any differences in the responses

based on the version. Note that all the Figures are

in the Appendix.

3.1 When ChatGPT 3.5 is used to Analyse

the Data

Analyst: (After copying the data set in text format

into the chat environment) check this data out.

How would it be displayed in R studio?

ChatGPT’s Response:Fig. (Appendix)

Comment: The ChatGPT got this instruction right.

It was even impressive that it had to put the data

inside a data frame when asked how it would be

displayed in R studio.

Analyst: What would the output look like?

ChatGPT’s Response: Fig. (Appendix)

Comment: When the researchers scrolled down,

they discovered some parts of the output had been

cut off to save space. Additionally, ChatGPT could

only generate a few parts of the output, even

though there were 30 observations; it only

produced 24 of them. This is the first limitation

noticed. Therefore, the ChatGPT may produce an

incomplete output when not accurately supervised

during use.

Analyst: Alright. Having done that, could you

suggest a statistical tool that could be considered

appropriate for analysing the data?

ChatGPT’s Response: Fig. 1a and 3b (Appendix)

Comment: A lot of unnecessary comments by

ChatGPT have been eliminated. The ChatGPT

seems to have a good glimpse of the nature of the

data. It even gave a hands-on code useful for

performing multiple linear regression using the R

studio. It however never suggested that

assumptions of multiple linear regression be tested

first before carrying out regression analysis. This is

another limitation of ChatGPT; being able to

decipher what tool to use and not being able to take

precautions on using such a tool.

Analyst: (It’s now being asked to perform multiple

linear regression on the data.) Since the multiple

linear regression could be appropriate, after taking

all necessary measures into consideration, proceed

with performing multiple regression on the data.

ChatGPT’s Response: Fig. (Appendix)

WSEAS TRANSACTIONS on COMPUTERS

DOI: 10.37394/23205.2024.23.3

Abiola Timothy Owolabi, Oluwaseyi Oluwadamilare Okunlola,

Emmanuel Taiwo Adewuyi,

Janet Iyabo Idowu, Olasunkanmi James Oladapo

E-ISSN: 2224-2872

Volume 23, 2024

Comment: There seems to be a “going back-and-

forth” in the output generated by ChatGPT. It failed

to carry out multiple regression analysis on the

data. Also, it encountered difficulties in completing

the code when attempting to place the data inside a

"data frame" in R Studio. This is a failure on the

part of the chatbot.

Analyst: Were you not trained to always test

assumptions before performing any analysis,

especially when using parametric test statistics?

ChatGPT’s Response: Fig. (Appendix)

Comment: The bot acknowledged the importance

of testing assumptions before data analysis. It

compiled a list of assumptions that needed to be

tested before performing multiple linear regression

analysis. However, it failed to carry out these

assumption tests and proceeded directly to conduct

the regression analysis with the datasets. This is

another limitation of ChatGPT; it could be

misleading when the person using it does not have

adequate knowledge of what they intend to use it

for. The Ordinary Least Square Estimator (OLSE)

is the Best Linear Unbiased Estimator (BLUE).

Nevertheless, in the presence of multicollinearity,

the OLSE can lead to regression coefficients with

incorrect signs, considerably large absolute values,

and extensive standard errors, [20], [21]. It means

using the OLSE like the ChatGPT in the presence

of multicollinearity will give a misleading result.

Further chats with ChatGPT about its inefficiency

resulted in some arguments. Here is the final

submission of the bot: Fig. (Appendix)

Comment: ChatGPT explicitly acknowledges that

it lacks personal qualities or capabilities.

Consulting with a qualified human data analyst or

Statistician is recommended for anyone seeking to

perform data analysis.

Analyst: Okay. Could you recall the data you put

inside the data frame?

ChatGPT’s Response: Fig. (Appendix)

Comment: It’s pretty impressive that the bot can

still recall the data inside a “data frame” after some

conversations.

Analyst: Alright. Now, test the assumptions of

multiple linear regression on the data. You have to

generate the code in R studio and the output of the

code as well. Also, write a brief note on the

outcome of each test.

ChatGPT’s Response: Fig. (Appendix)

Comment: The ChatGPT, instead of testing the

assumptions, listed and explained the underlying

assumptions of MLR. After further prompting, it

later generates the codes as seen in the next

response of the ChatGPT: Fig. (Appendix)

Looking at the response in Fig. (Appendix), it can

be realized that ChatGPT failed to write out codes

for testing the assumptions of the Multiple Linear

Regression (MLR) model. It still repeated the cycle

of putting the data inside a data frame.

Analyst: Now perform those tests on the data using

the R studio.

ChatGPT’s response: Fig. (Appendix)

Comment: In certain instances, ChatGPT has been

requested to place a dataset inside a data frame, run

a code, and successfully generate an output.

However, in a new development, ChatGPT raises

eyebrows about its inability to execute codes. Upon

closer examination of the output it did produce, it

becomes evident that complete codes have not been

written to test all the assumptions of the multiple

linear regression (MLR) model.

Analyst: Complete the rest from where you

stopped.

ChatGPT’s Response: Fig. (Appendix)

Comment: In this response, ChatGPT has provided

the complete codes for testing assumptions for the

MLR model. However, it's worth noting that

ChatGPT may not have been able to generate these

codes without prompting from the data analyst.

Furthermore, none of the codes have been executed

yet. This raises a crucial point: if researchers lack

the necessary knowledge for data analysis, the bot

could execute an incomplete code, resulting in

erroneous results.

Analyst: Run those codes in R studio, generate an

output, and write a brief note indicating whether or

not an assumption has been violated.

ChatGPT’s Response: Fig. (Appendix)

Comment: Up there, ChatGPT is lamenting its

limitation in running and executing codes. Another

WSEAS TRANSACTIONS on COMPUTERS

DOI: 10.37394/23205.2024.23.3

Abiola Timothy Owolabi, Oluwaseyi Oluwadamilare Okunlola,

Emmanuel Taiwo Adewuyi,

Janet Iyabo Idowu, Olasunkanmi James Oladapo

E-ISSN: 2224-2872

Volume 23, 2024

limitation of the ChatGPT is being noticed here. It

is worth noting that all the outputs generated thus

far by ChatGPT have been obtained through the

collaborative efforts of the data analyst, who assists

ChatGPT by providing contextual knowledge for

each chat.

After several attempts to get ChatGPT to execute

codes for testing of MLR model assumptions, the

final response by ChatGPT is found in Figure 13

(Appendix).

Note: Some parts were repeated in the response and

have hence been cut off. It could also be observed

from ChatGPT’s response that the bot lacks the

capability to run codes and generate textual

outputs.

3.2 When ChatGPT 4.0 is used to Analyse

the Data

Researchers undertook an investigation to assess

the capability of ChatGPT 4.0 for independent data

analysis without the involvement of human

analysts. This was prompted by the belief that

ChatGPT 4.0 possesses greater robustness and

intelligence compared to ChatGPT-3.5. The

conversations between the researchers and

ChatGPT 4.0 are outlined

as follows:

Analyst: Check this data out: Table 2. Economeric

Data Put Inside Table by ChatGPT 4 (Appendix)

Do these on the data

1) Put the data inside a table

2) What statistical tool is appropriate for analysing

the data?

3) Use that tool to analyse the data.

4) Are there any assumptions that must be met

before the statistical tool can be used?

5) Test those assumptions.

6) Why did you employ the statistical tool you used

in analysing the data?

7) Generate output for the test of assumptions and

the analysis you conducted.

8) Write a brief note on steps 1 to 7.

ChatGPT-4 Response:

ChatGPT responded by putting the data in Table

format (Table 2. Economeric Data Put Inside Table

by ChatGPT 4, Appendix). It continued its

discussions as follows:

There are several statistical tools that can

be used to analyze this data, depending on the

research question and the nature of the data. One

possible tool is multiple linear regression, which

can be used to model the relationship between a

dependent variable (y) and multiple independent

variables (x_1,x_2,x_3). This tool can help

determine if there is a significant relationship

between the independent variables and the

dependent variable. It can also provide information

about the strength and direction of those

relationships.

Before using multiple linear regression, several

assumptions must be met:

- Linearity: A linear relationship between the

independent and dependent variables should

exist.

- Independence: The observations should be

independent of each other.

- Normality: The residuals (errors) should be

normally distributed.

- Homoscedasticity: The variance of the residuals

should be constant across the range of the

independent variables.

These assumptions can be tested using various

statistical tests and visual methods such as scatter

plots, histograms, and Q-Q plots.

I employed multiple linear regression because

it is a widely used statistical tool for analyzing the

relationship between dependent and multiple

independent variables. It can provide valuable

information about the nature of those relationships

and can help answer research questions about the

data.

I'm sorry, but I am not able to generate an

output for the test of assumptions and the analysis

as I am an AI language model and do not have the

capability to run statistical tests or generate visual

outputs. However, you can use statistical software

such as R or SPSS to conduct these tests and

generate outputs.

In summary, I have put the data into a table

format, suggested multiple linear regression as a

possible statistical tool for analyzing the data,

explained the assumptions that must be met before

using this tool, and provided a rationale for why

this tool was employed. However, I am not able to

generate outputs for the test of assumptions or the

analysis. Is there anything else you would like to

know?

Comment: Based on the response of ChatGPT-4.0,

it can be concluded that its answers align with

those of ChatGPT-3.5. Therefore, both versions of

the Chabot exhibit the same limitations.

WSEAS TRANSACTIONS on COMPUTERS

DOI: 10.37394/23205.2024.23.3

Abiola Timothy Owolabi, Oluwaseyi Oluwadamilare Okunlola,

Emmanuel Taiwo Adewuyi,

Janet Iyabo Idowu, Olasunkanmi James Oladapo

E-ISSN: 2224-2872

Volume 23, 2024

4 Discussions

The ChatGPT is a fantastic tool that anyone who

understands its use will find helpful. However, it is

not without its limitations. Its use can only be

productive when its user has full knowledge of

what they intend to achieve with it.

In this study, ChatGPT’s prowess in data analysis

was examined. Using ChatGPT 3.5, the analyst

must make several promptings before it can

understand instructions, and at times, it generates

wrong outputs and misleading results. However,

ChatGPT 4.0 can generate the correct response.

When asked to analyse data, a lot of discrepancies

were noticed. The outputs it generated were mostly

erroneous; it had to be queried and prompted

severally before it could generate the correct

outputs. The ChatGPT itself submitted that the

service of a human data analyst would be required

to validate any result it produces. This is because at

some points in time, after several promptings and

queries, it could still not respond correctly. When

asked to generate codes to test assumptions of the

MLR model, it took a lot of effort to get it to

generate the needed codes, the codes it could not

even execute. How about writing a conclusion? It’s

a no for ChatGPT.

The peak of our findings in this study is the

submission of ChatGPT with respect to the need for

human Statistical analysts. In its words, after a

query due to its failure to perform a task as

expected, it responded, ‘I can offer guidance on

various topics, including data analysis, but it is

important to note that the final responsibility for

data analysis and interpretation rests with the user.

It’s always recommended to consult a qualified

human data analyst or Statistician to ensure

accurate and reliable results for complex data

analysis tasks.” Therefore, we may conclude that

ChatGPT will not be able to replace human data

analysts. The tool may come handy to data analysts

when stuck on what to do but should never be

considered an authority in statistical decisions.

Although ChatGPT may not be efficient for

data analysis, it can still be helpful by offering

valuable suggestions and writing codes. However,

it is essential that the analyst thoroughly checks

these suggestions and codes before adopting them.

Human data analysts need not fear losing their jobs

anytime soon; instead, they should focus on

enhancing their data analysis knowledge. This will

enable them to discern the correct approaches from

the incorrect ones, ensuring their continued

relevance and success in their field. This is

consistent with the submission of Ausat et al.

(2023), who submitted that ChatGPT can be used

as a tool in the learning process and not a substitute

for the role of the teacher in the classroom.

The ChatGPT may appear to be a valuable tool

for data analysis. However, it is referred to as a tool

for a reason, indicating that only individuals with

mastery of the tool can utilize it for optimal results.

It is undeniable that GPT is faster and can operate

at incredible speed when analyzing data compared

to humans.

Nevertheless, it remains reliant on humans as it

can only generate codes and offer suggestions for

data analysis, unable to perform any data analysis

independently. This signifies that human data

analysts are still irreplaceable in the field of data

analysis. Sometimes, ChatGPT can even produce

erroneous codes and provide incorrect suggestions

when asked for help. Data analysis with the tool

will always require the service of an expert in most

cases, [22].

5 Conclusion

This study has established that the role of ChatGPT

in data analysis is incredibly crucial, serving as a

valuable tool for knowledgeable data analysts. It

provides a natural language interface, making it

easier for analysts to ask questions, aiding in data

exploration, cleaning, and pre-processing.

Additionally, ChatGPT can explain complex

concepts in data analysis and provide information

in domain-specific knowledge.

However, to maximize its potential in data

analysis and interpretation for decision-making in

various domains of life, the expertise of a

knowledgeable and experienced Statistical Analyst

is necessary. This study has also revealed the

weaknesses and limitations of ChatGPT,

particularly when used by novices in the field of

Statistics.

The requirement for statistical knowledge,

critical thinking, ethical awareness, continuous

learning, and effective communication skills to

convey complex findings in a clear and

understandable manner to non-technical

stakeholders in data analysis can only be fulfilled

by human analysts. Therefore, we can infer that the

introduction of ChatGPT has enhanced the

efficiency and speed of a Data Analyst's job,

especially for those with a solid grasp of theoretical

statistics. This tool simplifies the analysis process

for them, providing a valuable resource when they

encounter challenges in statistical analysis.

WSEAS TRANSACTIONS on COMPUTERS

DOI: 10.37394/23205.2024.23.3

Abiola Timothy Owolabi, Oluwaseyi Oluwadamilare Okunlola,

Emmanuel Taiwo Adewuyi,

Janet Iyabo Idowu, Olasunkanmi James Oladapo

E-ISSN: 2224-2872

Volume 23, 2024

References:

[1]

C. McClelland, “The Impact of Artificial

Intelligence - Widespread Job Losses,” 30

January 2023. [Online].

https://www.iotforall.com/impact-of-

artificial-intelligence-job-losses (Accessed

Date: July 22, 2023).

[2]

Savelka, J.; Agarwal, A.; Bogart, C.; Song,

Y. and Sakr, M., “Can generative pre-trained

transformers (GPT) pass assessments in

higher education programming courses?”

arXiv preprint, p. arXiv:2303.09325, 2023,

https://doi.org/10.1145/3587102.3588792

[3]

Wamba, S. F., Bawack, R. E., Guthrie, C.,

Queiroz, M. M., & Carillo, K. D., “Are we

preparing for a good AI society? A

bibliometric review and research agenda,”

Technological Forecasting and Social

Change, p. 120482, 2020,

https://doi.org/10.2139/ssrn.3735322.

[4]

OpenAI, “ChatGPT,” Open AI, 2022.

[Online], https://chat.openai.com (Accessed

Date: March 31, 2024).

[5]

Adamopoulou, E., & Moussiades, L., An

overview of chatbot technology, Switzerland:

Springer Nature, 2020, pp. 373 - 383.

[6]

Fauzi, F.; Tuhuteru, L.; Sampe, F.; Ausat, A.

M. A. & Hatta, H. R., “Analysing the Role

of ChatGPT in Improving Student

Productivity in Higher Education,” J. Educ.,

vol. 5, no. 4, p. 14886–14891, 2023,

https://doi.org/10.31004/joe.v5i4.2563.

[7]

Ausat,A. M. A.; Massang, B.; Efendi, M.;

Nofirman; Riady, Y., “Can Chat GPT

Replace the Role of the Teacher in the

Classroom: A Fundamental Analysis,”

Journal on Education, vol. 5, no. 4, pp.

16100-16106, 2023,

https://doi.org/10.31004/joe.v5i4.2745.

[8]

A. Iskender, “Holy or Unholy? Interview

with Open AI's ChatGPT,” European

Journal of Tourism Research, vol. 34, no.

3414, 2023,

https://doi.org/10.54055/ejtr.v34i.3169.

[9]

U. Bhosale, “9 Things Researchers

Shouldn’t Expect From ChatGPT: Setting

realistic boundaries.,” Enago Academy, 8

June 2023, [Online].

https://www.enago.com/academy/chatgpt-

cannot-do-for-researchers (Accessed Date:

July 22, 2023).

[10]

S. O’Connor, “ChatGPT. Open artificial

intelligence platforms in nursing education:

Tools for academic progress or abuse?”

Nurse Educ Pract. , vol. 66:103537, p.

Medline:36549229, 2023,

https://doi.org/10.1016/j.nepr.2022.103537.

[11]

S. Huh, “Are ChatGPT’s knowledge and

interpretation ability comparable to those of

medical students in Korea for taking a

parasitology examination?: a descriptive

study,” J Educ Eval Health Prof., vol. 20,

no. 1, 2023,

https://doi.org/10.3352/jeehp.2023.20.01.

[12]

B. Kato, “AI stole my job — pay is down

90%, and my family is suffering,” 2 8 2023,

[Online]. https://nypost.com/2023/08/02/ai-

stole-my-job-pay-is-down-90-and-my-

family-is-suffering/ (Accessed Date: August

4, 2023).

[13]

Idowu, J.I., Oladapo, O. J., Owolabi, A. T.,

Ayinde, K. and Akinmoju, O., “Combating

Multicollinearity: A new two-parameter

Approach,” Journal of Quantitative

Sciences, vol. 5, no. 1, 2023,

https://doi.org/10.51541/nicel.1084768.

[14]

Owolabi, A. T., Ayinde, K. and Alabi, O. O.,

“A Modified Two Parameter Estimator with

Different Forms of Biasing Parameters in the

Linear Regression Model,” African Scientific

Reports, vol. 1, p. 212–228, 2022,

https://doi.org/10.46481/asr.2022.1.3.62.

[15]

Oladapo, O.J., Owolabi, A.T., Idowu, J.I.

and Ayinde, K. , “A New Modified Liu

Ridge-Type Estimator for the Linear

Regression Model: Simulation and

Application,” Int J Clin Biostat Biom, vol. 8,

no. 048, 2022,

https://doi.org/10.23937/2469-5831/1510048

[16]

Ayinde, K.; Lukman, A.F.; Samuel, O.O.; &

Ajiboye, S.A., “Some new adjusted ridge

estimators of linear regression model,” Int. J.

Civ. Eng. Technol., p. 2838-2852, 2018.

[17]

Lukman, A. F., Arashi, M. and Prokaj, V.,

“Robust biased estimators for Poisson

regression model: Simulation and

applications,” Concurrency and

Computation: Practice and Experience, vol.

35, p. e7594, 2023,

https://doi.org/10.1002/cpe.7594.

[18]

R. Cook, “Detection of Influential

Observations in Linear Regression,”

Technometrics, 1977.

[19]

Paul, S.R. and Fung, K. Y., “A generalized

extreme studentized residual multiple outlier

WSEAS TRANSACTIONS on COMPUTERS

DOI: 10.37394/23205.2024.23.3

Abiola Timothy Owolabi, Oluwaseyi Oluwadamilare Okunlola,

Emmanuel Taiwo Adewuyi,

Janet Iyabo Idowu, Olasunkanmi James Oladapo

E-ISSN: 2224-2872

Volume 23, 2024

detection procedure in Linear Regression,”

Technometrics, vol. 33, pp. 339-348, 1991,

https://doi.org/10.1080/00401706.1991.1048

4839.

[20]

Lukman, A.F. and Ayinde, K., “Review and

classifications of the ridge parameter

estimation techniques,” Hacettepe Math

Stat., vol. 46, no. 5, p. 953-967,

2017,https://doi.org/10.15672/hjms.2018156

71.

[21]

Owolabi, A. T., Ayinde, K., Idowu, J. I.,

Oladapo, O. J., & Lukman, A. F., “A New

Two-Parameter Estimator in the Linear

Regression Model with Correlated

Regressors,” Journal of Statistics

Applications & Probability, vol. 11, pp. 499-

512, 2022,

https://doi.org/10.18576/jsap/110211.

[22]

A. Sakamoto, “ChatGPT Data Analysis vs.

Human Analysis: Which One Should You

Choose?,” Kanaries, 31 07 2023, [Online].

https://docs.kanaries.net/articles/chatgpt-

data-analysis-human (Accessed Date:

August 4, 2023).

WSEAS TRANSACTIONS on COMPUTERS

DOI: 10.37394/23205.2024.23.3

Abiola Timothy Owolabi, Oluwaseyi Oluwadamilare Okunlola,

Emmanuel Taiwo Adewuyi,

Janet Iyabo Idowu, Olasunkanmi James Oladapo

E-ISSN: 2224-2872

Volume 23, 2024

APPENDIX

Table 1. Identification of outliers in the datasets

dfb.1_

dfb.x1

dfb.x2

dfb.x3

dffit

cov.r

cook.d

hat

-0.056

-0.045

0.026

0.046

-0.067

1.300

0.001

0.106

-0.460

-0.183

0.210

0.185

-0.471

0.709

0.050

0.056

0.060

0.243

-1.171

-0.238

-1.308

0.457

0.335

0.173

-0.286

0.137

0.148

-0.146

-0.584

0.739

0.077

0.085

-0.311

0.084

0.152

-0.088

-0.419

0.757

0.040

0.051

-0.206

0.042

0.085

-0.037

-0.334

1.235

0.028

0.146

0.088

-0.084

-0.033

0.080

0.196

1.441

0.010

0.207

0.060

0.022

-0.030

-0.021

0.082

1.235

0.002

0.067

-0.034

0.058

0.015

-0.058

-0.078

1.279

0.002

0.094

0.046

0.041

-0.024

-0.040

0.077

1.316

0.002

0.118

0.055

0.032

-0.025

-0.032

0.066

1.292

0.001

0.100

0.060

0.075

-0.030

-0.074

0.088

1.382

0.002

0.159

-0.029

-0.044

-0.147

0.047

-0.208

1.485

0.011

0.230

-0.012

-0.101

-0.234

0.100

-0.306

1.387

0.024

0.204

0.007

0.008

-0.003

-0.008

0.010

1.445

0.000

0.191

-0.127

0.069

0.061

-0.071

-0.189

1.125

0.009

0.056

-0.069

-0.040

0.034

0.039

-0.084

1.231

0.002

0.065

0.276

0.089

-0.114

-0.096

0.427

1.271

0.046

0.187

0.062

0.453

0.610

-0.448

0.934

1.091

0.206

0.262

0.242

-0.141

-0.125

0.149

0.508

0.846

0.060

0.086

0.009

-0.244

0.572

0.232

0.794

1.221

0.153

0.267

-0.109

-0.279

0.427

0.282

0.558

1.430

0.078

0.279

0.207

-0.140

-0.088

0.137

0.306

1.049

0.023

0.073

0.068

-0.165

-0.027

0.164

0.204

1.323

0.011

0.149

0.090

-0.121

-0.045

0.124

0.220

1.196

0.012

0.092

0.134

-0.009

-0.062

0.009

0.142

1.136

0.005

0.042

0.029

-0.026

-0.015

0.027

0.064

1.272

0.001

0.087

-0.031

0.130

0.015

-0.131

-0.157

1.395

0.006

0.177

0.169

0.157

-0.088

-0.153

0.286

1.220

0.021

0.125

0.083

0.014

-0.042

-0.013

0.122

1.212

0.004

0.066

NOTE: The bolded observations are outliers.

WSEAS TRANSACTIONS on COMPUTERS

DOI: 10.37394/23205.2024.23.3

Abiola Timothy Owolabi, Oluwaseyi Oluwadamilare Okunlola,

Emmanuel Taiwo Adewuyi,

Janet Iyabo Idowu, Olasunkanmi James Oladapo

E-ISSN: 2224-2872

Volume 23, 2024

Table 2. Economeric Data Put Inside Table by ChatGPT 4

0.355772049

-0.366547941

-1.182745938

-0.91417413

-0.416631134

-0.25264793

-0.206310012

-0.371176738

0.768840797

-0.535158825

13.49243216

-0.299338951

1.870846452

0.778642334

1.120420843

1.188066458

1.538532252

0.663850856

-0.066041848

1.000284307

-0.949063094

-1.107773201

-1.903406801

-0.747583634

-1.216966558

-1.972597717

-1.288179357

-1.125329398

1.960095769

0.645110609

0.03686517

0.593879336

0.884327286

-0.575699024

-0.008114294

0.378645915

2.25594158

0.992529211

0.820315829

0.561067384

0.455959839

-0.205369354

-0.212007131

-0.519583762

1.294878123

0.706875387

0.045052258

-0.246070919

0.407722362

-1.039792628

11.71049892

-1.392955829

3.437947651

1.070833737

14.01180889

0.54517308

-0.094945274

-0.246727249

-0.991681436

-1.219339862

2.172972119

0.545802769

0.494296884

1.018028015

1.152682348

0.178637641

0.696243381

-0.050153627

-0.534216222

-1.577309852

-1.623539592

-1.679847471

4.625167052

1.328776544

13.63763941

0.466952523

5.36849768

1.225683711

1.300765358

1.678838253

2.455415364

-1.412064924

11.68088058

-0.893731385

4.887618937

0.063436956

13.87780516

0.918925503

1.992924213

-0.711888317

0.17855623

-0.106977664

1.463271696

-1.207861548

0.647646888

0.028174873

3.525824643

0.2803768

1.052345338

1.023760812

2.373526151

-0.093445044

0.41218895

0.138662914

3.446087155

0.636232781

0.542666176

1.223515765

3.276991747

0.287532367

0.899982

1.635599667

3.225472496

1.085633778

0.848451595

0.613920679

3.0287778

0.745206701

0.982665575

0.823590469

WSEAS TRANSACTIONS on COMPUTERS

DOI: 10.37394/23205.2024.23.3

Abiola Timothy Owolabi, Oluwaseyi Oluwadamilare Okunlola,

Emmanuel Taiwo Adewuyi,

Janet Iyabo Idowu, Olasunkanmi James Oladapo

E-ISSN: 2224-2872

Volume 23, 2024

Fig. 1: ChatGPT’s Response to the first question of the Analyst

Fig. 2: ChatGPT’s Response to the second question of the Analyst

WSEAS TRANSACTIONS on COMPUTERS

DOI: 10.37394/23205.2024.23.3

Abiola Timothy Owolabi, Oluwaseyi Oluwadamilare Okunlola,

Emmanuel Taiwo Adewuyi,

Janet Iyabo Idowu, Olasunkanmi James Oladapo

E-ISSN: 2224-2872

Volume 23, 2024

Fig. 1a: ChatGPT’s Response to the third question of the Analyst

Fig. 3b: Part of ChatGPT’s Response to the third question of the Analyst

Fig. 4: ChatGPT’s Response to the fourth question of the Analyst

WSEAS TRANSACTIONS on COMPUTERS

DOI: 10.37394/23205.2024.23.3

Abiola Timothy Owolabi, Oluwaseyi Oluwadamilare Okunlola,

Emmanuel Taiwo Adewuyi,

Janet Iyabo Idowu, Olasunkanmi James Oladapo

E-ISSN: 2224-2872

Volume 23, 2024

Fig. 5: ChatGPT’s Response to the fifth question of the Analyst

Fig. 6: ChatGPT’s Response to the sixth question of the Analyst

WSEAS TRANSACTIONS on COMPUTERS

DOI: 10.37394/23205.2024.23.3

Abiola Timothy Owolabi, Oluwaseyi Oluwadamilare Okunlola,

Emmanuel Taiwo Adewuyi,

Janet Iyabo Idowu, Olasunkanmi James Oladapo

E-ISSN: 2224-2872

Volume 23, 2024

Fig. 7: ChatGPT’s Response to the seventh question of the Analyst

Fig. 8: ChatGPT’s Response to the eighth question of the Analyst

WSEAS TRANSACTIONS on COMPUTERS

DOI: 10.37394/23205.2024.23.3

Abiola Timothy Owolabi, Oluwaseyi Oluwadamilare Okunlola,

Emmanuel Taiwo Adewuyi,

Janet Iyabo Idowu, Olasunkanmi James Oladapo

E-ISSN: 2224-2872

Volume 23, 2024

Fig. 9: ChatGPT’s Response to the ninth question of the Analyst

Fig. 10: ChatGPT’s Response to the tenth question of the Analyst

WSEAS TRANSACTIONS on COMPUTERS

DOI: 10.37394/23205.2024.23.3

Abiola Timothy Owolabi, Oluwaseyi Oluwadamilare Okunlola,

Emmanuel Taiwo Adewuyi,

Janet Iyabo Idowu, Olasunkanmi James Oladapo

E-ISSN: 2224-2872

Volume 23, 2024

Fig. 11: ChatGPT’s Response to the eleventh question of the Analyst

WSEAS TRANSACTIONS on COMPUTERS

DOI: 10.37394/23205.2024.23.3

Abiola Timothy Owolabi, Oluwaseyi Oluwadamilare Okunlola,

Emmanuel Taiwo Adewuyi,

Janet Iyabo Idowu, Olasunkanmi James Oladapo

E-ISSN: 2224-2872

Volume 23, 2024

Fig. 12: ChatGPT’s Response to the twelfth question of the Analyst

Fig. 13: ChatGPT’s Response to the thirteenth question of the Analyst

Contribution of Individual Authors to the

Creation of a Scientific Article (Ghostwriting

Policy)

The authors equally contributed in the present

research, at all stages from the formulation of the

problem to the final findings and solution.

Sources of Funding for Research Presented in a

Scientific Article or Scientific Article Itself

No funding was received for conducting this study.

Conflict of Interest

The authors have no conflicts of interest to declare.

Creative Commons Attribution License 4.0

(Attribution 4.0 International, CC BY 4.0)

This article is published under the terms of the

Creative Commons Attribution License 4.0

https://creativecommons.org/licenses/by/4.0/deed.e

n_US

WSEAS TRANSACTIONS on COMPUTERS

DOI: 10.37394/23205.2024.23.3

Abiola Timothy Owolabi, Oluwaseyi Oluwadamilare Okunlola,

Emmanuel Taiwo Adewuyi,

Janet Iyabo Idowu, Olasunkanmi James Oladapo

E-ISSN: 2224-2872

Volume 23, 2024