Data Analytics Tasks to Finish in 4 Weeks

1. The 4-Week Junior Analyst Sprint Overview

Many aspiring data analysts spend months watching passive video tutorials on syntax without ever completing an end-to-end analytical project. In corporate analytics teams, your value is determined by how efficiently you can intake messy raw data, validate assumptions, query relational databases, and deliver visual answers that executive decision-makers understand.

A structured four-week internship sprint simulates a corporate business intelligence engagement. By allocating 15 to 20 hours per week across four distinct project phases, you develop muscle memory across the entire data lifecycle.

This guide outlines the exact tasks, deliverables, and technical criteria evaluated by industry mentors during an intensive four-week analytics internship.

2. Week 1: Data Hygiene & Pandas Cleaning Pipelines

Real-world business data is chaotic: customer phone numbers contain formatting errors, timestamps use mismatched timezones, currency values are stored as strings with dollar signs, and duplicate records pollute revenue calculations. Your first week is dedicated to mastering data hygiene.

Using Python and the Pandas library (or PostgreSQL ETL scripts), ingest a raw dataset containing at least 10,000 transaction records. Construct a reproducible cleaning pipeline with documented assumptions.

# Example: Defensive data cleaning pipeline in Pandas
import pandas as pd

def clean_transactions(df: pd.DataFrame) -> pd.DataFrame:
    # Deduplicate on primary transaction identifier
    cleaned = df.drop_duplicates(subset=['transaction_id']).copy()
    
    # Normalize currency strings into numeric floats
    cleaned['revenue'] = (
        cleaned['revenue']
        .astype(str)
        .str.replace(r'[^0-9.]', '', regex=True)
        .astype(float)
    )
    
    # Parse timestamps with timezone normalization
    cleaned['transacted_at'] = pd.to_datetime(cleaned['transacted_at'], utc=True)
    
    # Assert no negative revenue values
    assert (cleaned['revenue'] >= 0).all(), "Negative revenue values detected!"
    return cleaned

3. Week 2: Advanced SQL Modeling & Exploratory Analysis

In Week 2, load your sanitized dataset into a relational database such as PostgreSQL or SQLite. The focus shifts to advanced SQL querying: identifying statistical trends, cohort behavior, and seasonal variations.

Junior analysts often rely on simple GROUP BY queries, but enterprise roles require proficiency with Window Functions and Common Table Expressions (CTEs).

Document each query with a brief explanation of the business hypothesis tested and the resulting metric conclusion. Submit your SQL queries in a well-organized `.sql` script file with syntax highlighting.

4. Week 3: Interactive Executive Dashboards in Tableau or PowerBI

Data that cannot be easily understood by non-technical stakeholders has zero commercial value. In Week 3, connect your sanitized dataset to an industry-standard Business Intelligence platform: Tableau Public, Microsoft PowerBI, or Streamlit.

Your deliverable is a single-screen executive dashboard designed for a Chief Revenue Officer or Product Manager. The dashboard must balance high-level summary cards with granular interactive exploration tools.

5. Week 4: Business Insights, Storytelling & Feature Importances

The final week bridges the gap between data computation and strategic decision-making. You will write a 2-page Executive Insights Brief summarizing your empirical findings.

Structure your report using the Pyramid Principle: state your primary conclusion and recommended action in the first paragraph, supported by secondary evidence and quantitative charts.

If you pursue the optional stretch milestone, train a baseline classification model using Scikit-Learn (such as Logistic Regression or a Random Forest) to predict customer churn. Calculate Precision, Recall, and ROC-AUC curves, and map feature importances back to operational business fixes.

6. How to Package Your Analytics Portfolio for Hiring Managers

To maximize your four-week investment, package your deliverables into an unassailable public repository on GitHub: