How to become a junior data analyst: what to learn first

A data analyst turns data into answers for the business: which product sells the most?, why did sales drop in March?, which customers stopped buying? Analysts are hired for their questions and conclusions, not just their charts.

1. Data analyst vs. data scientist

• A data analyst answers what happened and why, with spreadsheets, SQL, Python and dashboards.
• A data scientist builds models to predict what will happen, with more statistics and machine learning.

To get started, the analyst path is more direct: good queries and a good story already add value.

2. What to learn and in what order

1. Spreadsheets: cleaning (duplicates, extra spaces, empty cells), SUMIF, XLOOKUP and pivot tables. And knowing when to use the mean or the median.
2. Basic SQL: SELECT, WHERE, ORDER BY, aggregations and GROUP BY with HAVING.
3. SQL for business: JOIN across several tables, LEFT JOIN to find what's missing, subqueries and CASE WHEN to segment.
4. Python with pandas in Google Colab: load a CSV, explore, clean and group.
5. Visualization and dashboards: pick the right chart, define KPIs and build a dashboard in Looker Studio or Power BI.
6. Data storytelling: context, conflict, finding and recommendation.

3. A pandas example

import pandas as pd

sales = pd.read_csv("sales.csv")
sales["total"] = sales["price"] * sales["quantity"]
by_category = (sales.groupby("category")["total"]
               .sum()
               .sort_values(ascending=False))
print(by_category.head())

You can do the same with a pivot table or with GROUP BY in SQL. Knowing all three makes you flexible.

4. Portfolio projects

• A store's sales: the same analysis in a spreadsheet and in SQL, with HIGH, MEDIUM and LOW customer segments.
• Exploratory pandas notebook: a clean CSV, 5 questions answered and 3 charts with written conclusions.
• Dashboard with public data: Looker Studio or Power BI with a dataset from an open data portal or Kaggle, plus 5 slides with your recommendation.

5. Common mistakes

• Starting with the tool instead of the question. Write down what you want to know first.
• Charts without a conclusion. The chart title should say what it shows: "Online sales grew every month".
• Confusing correlation with causation. Two things rising together doesn't prove one causes the other.
• Not reading the dataset's documentation. Check where the data comes from, what each column means and its license.

If you've never coded, start with how to start programming from scratch. To go deeper into queries, read learn SQL from scratch.

Next step

Codea's free Data Analyst Jr route takes you in 6 weeks from spreadsheets to SQL, pandas and dashboards, ending with a final project on real public data reviewed by a mentor. Route content is in Spanish.