How to Write a Data Analysis Report

A data analysis report explains what question you asked of a dataset, how you prepared and analyzed it, what you found and how far the findings can be trusted. A reader should be able to repeat your analysis from the report alone.

This page sets out the standard structure, what goes in each section, how to present results and the mistakes that cost marks. It includes an example results paragraph, a short code sample and a checklist.

Report Structure at a Glance

Start with a Precise Question

A vague aim such as "analyze the sales data" leads to a vague report. Turn it into a question the data can answer: "Did average weekly sales differ between stores that ran the promotion and those that did not?"

State the question in the introduction and return to it in the discussion. Every figure and table in between should help answer it.

Describe the Data

• Source: who collected it, when and how, with a citation if it is public.
• Size: the number of rows (observations) and columns (variables).
• Variables: name, type (numeric, categorical, date) and units for each one you use.
• Unit of analysis: what one row represents, such as one patient, one sensor reading or one transaction.

A small table of variables with their units and a short description saves the reader from guessing what each column means.

Cleaning and Preparation

Report every change you made to the raw data. Cleaning decisions can change results, so a marker needs to see them.

• How many missing values each key variable had, and how you handled them (removed, imputed or kept).
• How you identified outliers and whether you kept them. Never drop a value only because it is inconvenient.
• Any transformations, such as log scales, unit conversions or new derived variables.
• How many rows remained after each step.

import pandas as pd

df = pd.read_csv("sales.csv")
print(df.shape)                    # rows, columns before cleaning
print(df.isna().sum())             # missing values per column
df = df.dropna(subset=["weekly_sales"])
print(df.shape)                    # rows remaining

Methods and Justification

Name each method and say why it suits the question and the data. "We used Welch's t-test because the two groups had unequal variances" is better than "a t-test was performed".

State the software and version, the significance level if you test hypotheses, and any settings that matter, such as the train-test split or random seed in a machine learning model.

Results and Figures

Present results in the order of your questions. Lead each paragraph with the finding, then give the numbers, then point to the figure or table.

• Label every axis with the variable name and unit.
• Give every figure a number and a caption that says what it shows.
• Report uncertainty: confidence intervals, standard errors or error bars, and say which.
• Round sensibly; three significant figures is usually plenty.

Interpretation and Limitations

Explain what the results mean for the question in plain language. Then be honest about what the analysis cannot show.

• Causation: observational data show association. Stores chose whether to run the promotion, so other differences may explain the gap.
• Data quality: missing values, measurement error or a short time window.
• Generality: whether the findings would hold for other regions, periods or populations.

Finish with a short, specific recommendation or next step, such as a randomized trial of the promotion.

Checklist Before You Submit

• The question is stated precisely and answered in the discussion.
• Data source, size, variables and units are described.
• Every cleaning step is reported with row counts.
• Each method is named and justified.
• Figures have numbered captions and labeled axes with units.
• Results include effect sizes or uncertainty, not just p-values.
• Limitations are specific, not generic.
• Code runs from start to finish on the raw data.

How STEM Donkey Helps with Data Analysis Reports

Send your dataset, the questions or brief, the required software (Python, R, SPSS, Excel or another) and the rubric. Deadlines run from 3 hours to 20 days.

A data analyst prepares a custom report written from scratch, with reproducible code, labeled figures and an honest interpretation. It is plagiarism-checked and meant for study and reference. We never invent or alter data.

Choose a Standard, Master or Elite writer, see the price before you pay, and get free revisions within the scope of your original request. Your identity is never shared with the writer.