Leveraging AI for Research Data Analysis: From Data Search to Generating Figures and Tables for Papers

When writing a paper or policy report, there are times when you need to examine a research topic numerically. For example, comparing housing cost burdens by region or identifying differences based on household type. Even if you find relevant public data, you might feel overwhelmed about which items to choose and how to analyze them.

Tlooto AI supports web data exploration and download, analysis according to researcher-specified methods, interpretation of results, and organization of figures and tables. It assists throughout the entire process, from data acquisition to finalizing conclusions for your manuscript, thereby reducing the time and effort required at each stage. Before analysis, define the questions you want to answer, then select appropriate data and methods.

Using a study comparing the housing cost burden of young households by region as an example, we will examine the process from data search to incorporating results into a paper.

Check Tlooto features at Hancom Addons

*The variables and analysis requests below are hypothetical examples to explain the process and do not represent actual analysis results.

Define Your Research Questions for Data Analysis

Before searching for data, write down what you want to know in a single sentence. A scope like “I want to analyze youth housing issues” is too broad and needs to be narrowed down. By refining it to “How does the housing cost burden of young households differ by region?”, it becomes clear that you need to define the survey regions, the scope of young households, and indicators to represent housing cost burden.

Even when addressing the same housing issue, the necessary data and analysis methods vary depending on the question. To determine “Did housing support programs reduce the burden on young households?”, simply comparing regional figures is insufficient. This is because you need to examine conditions for evaluating policy effectiveness, such as changes before and after program implementation and a comparison group.

In this article, we will focus on the first question: regional differences. Below your research question, list the subjects of analysis and the necessary items. If you want to compare young households, you might organize it by checking if household-level data is available and noting that you need criteria to distinguish young households, as well as housing cost, income, and regional information. Defining the scope in this way allows you to concretely organize the items to check during data search. If you find it difficult to determine the necessary items, present your research question to Tlooto and ask what data and variables you should examine. Referring to the suggested items can help you organize the conditions to check before finding data and establish criteria for selecting data that aligns with your research objectives.

Find Research Data and Verify Sources and Terms of Use

Once you have determined the necessary items, use Tlooto to find research data available on the web. Providing the region you are looking for, the survey period, and the required variables can help in exploring data relevant to your research objective. This allows you to examine the conditions needed for analysis along with the data name, making the criteria for selecting candidate data clearer. When comparing young households by region, you should first check if the data includes information to distinguish age or household type. Data that only provides overall regional averages makes it difficult to specifically identify the characteristics of young households.

After finding data, do not immediately proceed with analysis; read the guidelines provided by the data provider. You must confirm whether the data can be processed or entered into external services, and what conditions apply when sharing the results. If the public data includes copyrighted material, also check the scope of use, such as the KOGL type. You can also find related information in the Ministry of Interior and Safety’s Public Data Use Policy.

After downloading the data, record the providing institution, data name, survey year, and update date or version. You should also cross-reference whether the data on the description page matches the downloaded file. If an item description is provided with the file, you can use it to verify the units of values, survey subjects, and the meaning of codes.

If you are using personal information or non-public institutional data, you must first verify institutional regulations and obtain necessary approvals. Organizing the scope of use will be helpful not only when requesting analysis but also when incorporating results into papers or reports.

Define Variable Meanings and Analysis Methods

Once you have secured the data to be used, it’s time to confirm what each row and column in the file represents. The unit of analysis changes depending on whether a row represents an individual or a household. This basic unit that constitutes the data is called the ‘observational unit.’ If you are comparing young households by region, check if the data is at the household level and organize the meaning of the variables in each column.

Below are examples of variables that might be needed for this study. The actual classification criteria should follow the data provider’s documentation.

Hypothetical Variable Name	Meaning	Points to Check Before Analysis
region	Survey Region	Regional codes and administrative district criteria
survey_year	Survey Year	Whether the survey periods of the data to be compared are the same
householder_age	Householder Age	Whether to classify young households by householder age and the applicable age range
household_type	Household Type	Classification criteria for types such as single-person households
income_month	Monthly Household Income	Currency unit and pre-tax or post-tax basis
housing_cost_month	Monthly Housing Cost	Included costs and aggregation period
sample_weight	Sample Weight	Availability of weights and application guidelines

Even if variable names are the same, measurement criteria may differ. If income is an annual amount and housing cost is a monthly amount, they must be aligned to the same period. You also need to check whether housing costs include only monthly rent or other expenses like maintenance fees to ensure consistent comparison criteria.

Establish Criteria for Handling Empty or Uncalculable Values

After confirming the meaning of variables, check for missing values—data that is empty or omitted. If you are trying to calculate the proportion of income spent on housing, but some households have no income or zero income recorded, you cannot simply divide. You must decide how to handle such data based on your research objective and data characteristics, and record your reasoning.

During this process, also define your analysis methods concretely. When comparing regional burden levels, decide whether to look at the average, median, or the entire distribution, and verify if the method is appropriate for your data. If it’s difficult to choose a method, you can explain your research question and data characteristics to Tlooto and ask it to summarize methods for consideration and their application conditions. By comparing what each method can reveal and its application conditions, you can refine your analysis plan. If sample weights are provided, you must also check the application guidelines. Select the final analysis method after reviewing its suitability for your actual data and research question.

By organizing variables and processing criteria in this way, you can explain the tasks to Tlooto more concretely. Check the service’s scope of support to see if your prepared data and analysis methods can be used.

Explore Tlooto’s Usage Environment at Hancom Addons

Submit Analysis Conditions and Review Results

Once you have decided on the analysis method, submit the data and conditions together. When requesting methodology-based analysis, it’s better to specify the analysis subjects, variables, and calculation methods rather than simply asking, “Draw conclusions from this data.” This way, you can verify whether the analysis was performed according to your requested conditions after receiving the results.

Based on the information organized earlier, you can make the following request. For the brackets, insert your actual data and researcher-defined criteria:

I want to compare the housing cost burden of young households by region using [Data Name and Survey Year]. Please classify young households by [Classification Criteria] and use [Variable Name and Unit] for housing costs and income. Analyze according to [Analysis Method], [Criteria for Handling Missing Values and Households with Zero Income], and [Weight Application Criteria]. Provide a regional results table, along with the number of households included in the analysis, reasons for exclusion, and points to note in interpreting the results.

Upon receiving the results, first verify that the requested data and scope are correct. Examine the number of households included in the analysis and the reasons for exclusion, and cross-reference whether currency units or weights were applied according to the defined criteria. Keep the files used, request details, processing steps, and results together so that key calculations can be re-verified.

Record Changed Conditions and Reasons When Correcting Errors

While reviewing results, you might discover that currency units were applied incorrectly or missing values were mishandled. In such cases, you must record the error and the corrected conditions, then compare the re-analyzed results. Also, examine how the number of included households and the result values changed before and after correction.

If the results differ from expectations even without errors, that difference also requires explanation in the research. You must not arbitrarily change analysis conditions to achieve a desired conclusion. Review the reasons for unexpected results, but distinguish between facts confirmed by the data and the researcher’s interpretation.

For example, even if the housing cost burden in a specific region appears low, you cannot immediately attribute this to the effectiveness of housing support policies. This analysis aimed to compare regional differences. Explaining policy effects would require additional data and analysis design tailored for that purpose.

Organize Figures and Tables and Cross-Reference with Body Text

Once you have reviewed the analysis results, it’s time to organize them into figures or tables that are easy for the reader to understand. Tlooto also supports organizing analysis results into figures and tables suitable for papers and reports. Structure them according to your purpose; for instance, use graphs to show regional differences or distributions, and tables to present specific numerical values. By utilizing the figures and tables organized by Tlooto, you can reduce the burden of re-structuring analysis results to fit your manuscript.

When creating figures and tables for a paper, don’t just check for numerical accuracy; readers must also understand what was compared and under what conditions the analysis was performed. Review the following items, including titles, units, and sources:

Items to Review	What to check
Title	Does it clearly indicate the subject, period, and indicators compared?
Units and Axes	Are the units for amounts or percentages displayed, and are the axis ranges set appropriately?
Subjects of Analysis	Are the scope of households included in the analysis and exclusion criteria explained?
Analysis Conditions	Are the analysis method, handling of missing values, and application of weights recorded?
Data Source	Can the providing institution, data name, survey year, and version be identified?
Body Text Interpretation	Do the results observed in the figures or tables align with the explanations in the body text?

Narrowing the axis range of a graph can exaggerate small differences, and if percentage bases differ, direct comparison of values becomes difficult. Confirm that comparisons are made on the same basis, and if uncertainty is indicated (e.g., with error bars), explain what it signifies. If you are using figures or tables from other sources, you must separately verify their sources and terms of use.

Finally, insert the results into your Korean manuscript and cross-reference them with the body text again. Check if the titles, units, and annotations of the figures and tables are easy to read, and include the data source and analysis conditions. Since the method of transmitting response text and inserting figures or tables may differ, also confirm the actual file format and insertion method provided.

Research Data Analysis AI FAQ

Start by Examining a Single Piece of Data Relevant to Your Research Question

To utilize AI for research data analysis, you must first define what to check and under what conditions to analyze. Identify the parts of your current Korean manuscript that need numerical explanation, and then use Tlooto to find relevant public data. After verifying the source, terms of use, and meaning of variables, you can request analysis conditions, necessary results tables, and a draft explanation. By reflecting the reviewed results in your manuscript, continue from data exploration to organizing and writing the analysis content.

All Tlooto features can be found on the Tlooto website. For information on the Korean-language environment and how to get started, refer to the Hancom Addons guide.

Check Tlooto Getting Started Guide at Hancom Addons

Official Sources

• tlooto Official Website
• HanAddons Official Guide
• Ministry of Interior and Safety Public Data Use Policy: Scope of Use
• APA Style: Plagiarism