An end-to-end data analysis project analyzing Netflix's content catalog using Microsoft Excel, PostgreSQL, and Power BI. The project follows a real-world analyst workflow — raw data ingestion, cleaning, SQL-based analysis, and interactive dashboard reporting — addressing 15 business problems across content distribution, growth trends, genre patterns, director insights, geographic reach, and audience classification.
Netflix_Data_Analysis_Project/
│
├── Dashboard/
│ ├── netflix_dashboard.pbix
│ └── Dashboard_Images/
│ ├── Netflix Dashboard Home Page.png
│ ├── Page 1 (Content Distribution).png
│ ├── Page 2 (Growth Trends).png
│ ├── Page 3 (Content & Creators).png
│ └── Page 4 (Geographic Reach).png
│
├── data/
│ ├── Cleaned data/
│ │ ├── csv/
│ │ │ └── netflix_clean_data.csv
│ │ └── excel_file/
│ │ └── netflix_cleaning.xlsx
│ └── Raw data/
│ └── netflix_titles_raw_data.csv
│
├── result_set_csv_files/
│ ├── 1_content_distribution(1).csv
│ ├── 1_content_percentage(2).csv
│ ├── 2_cumilative_content(4).csv
│ ├── 2_growth_YOY(3).csv
│ ├── 2_yearly_content_type_distribution(Bonus).csv
│ ├── 3_top_genres_for_each_year(6).csv
│ ├── 3_top_genres(5).csv
│ ├── 4_top_directors_for_each_content_type(8).csv
│ ├── 4_top_directors_working_both_content_types(9).csv
│ ├── 4_top_directors(7).csv
│ ├── 5_top_countries_content_count(10).csv
│ ├── 5_top_countries_content_type_distribution(12).csv
│ ├── 5_top_countries_of_above_avg_content_count(11).csv
│ ├── 6_rating_by_audience_category_distribution(14).csv
│ ├── 6_ratings_by_audience_content_type_wise_distribution(15).csv
│ └── 6_ratings_wise_distribution(13).csv
│
├── Scripts/
│ ├── 0_db_&_table_creation_and_setup.sql
│ ├── 1_content_distribution.sql
│ ├── 2_growth_trends.sql
│ ├── 3_genre_analysis.sql
│ ├── 4_director_analysis.sql
│ ├── 5_country_analysis.sql
│ └── 6_rating_distribution.sql
│
└── README.md
| Property | Details |
|---|---|
| Source | Kaggle — Netflix Movies and TV Shows by Shivam Bansal |
| Raw File | netflix_titles_raw_data.csv |
| Total Records | ~8,807 rows (post-cleaning) |
| Columns | 12 — show_id, type, title, director, cast, country, date_added, release_year, rating, duration, listed_in, description |
Known data quality issues present in raw data:
- ~30% missing values in the
directorcolumn - Multi-value columns —
cast,country, andlisted_incontain multiple comma-separated values stored in single cells - Inconsistent date formats with leading/trailing spaces in
date_added - Misplaced values — duration values (
74 min,84 min) incorrectly stored in theratingcolumn - One fully corrupted row (row 8423) with all fields shifted due to an unescaped comma in the source CSV
- Non-standard rating values —
NRandURpresent alongside standard ratings
| Tool | Role in this project |
|---|---|
| Microsoft Excel 2019 | Data ingestion, cleaning, Removing duplicates, blank handling, exploratory pivot tables |
| PostgreSQL via pgAdmin 4 | Database creation, table schema setup, data reviewing after manipulating |
| VS Code | Importing cleaned data into database created in PostgreSQL, Writing and executing SQL analysis scripts, GitHub integration for local and remote repository management, README authoring and change tracking |
| DBeaver | Viewing table rows and data contents without the need for querying, used to save time during data validation and exploration |
| Power BI Desktop | Result-set import via Power Query, DAX measures, interactive 4-page dashboard |
| GitHub Copilot | Automation of pre-written SQL query's inline comments beside each query lines (mentioned in "--") which repeat across all scripts in Scripts/(explaining the role of all the individual lines of SQL code/query) |
| Claude AI | AI-assisted code review, project-documentation assistance, and project planning assistance to resolve bottlenecks on the independent plan — All the major planning & queries have been scripted independently(i.e without using these above mentioned AI tools) for practicing and sharpening the tool based knowledge & skills |
The raw CSV was imported into Excel using Data → Get & Transform → From Text/CSV with UTF-8 encoding to preserve special characters. All cleaning was performed on a working copy, keeping the original raw sheet untouched.
Key cleaning steps performed:
- Removed duplicates using
show_idas the unique identifier via Data → Remove Duplicates - Handled blank values — columns with few missing rows (
release_year,duration) were filled manually by cross-referencing the internet; high-volume blank columns (director,cast,country,listed_in) were bulk-filled using Filter → Blanks → Ctrl+Enter - Fixed date formatting — used
=--TRIM(cell)to strip leading/trailing spaces fromdate_addedwhile preserving the underlying date serial number, then reformatted to a consistent date display format - Corrected misplaced values — duration values found in the
ratingcolumn were moved toduration; affected rating cells set toUnknown - Standardized ratings —
NRandURreplaced withNot Ratedvia Find & Replace with Match Entire Cell Contents enabled - Handled corrupted row — row 8423 had all fields shifted due to an unescaped comma in the source CSV; fields were manually corrected; unrecoverable
show_idassigned placeholders9999 - DATE and INT columns — left as empty cells so PostgreSQL imports them as proper
NULLvalues without type errors - Exploratory pivot tables — built in a dedicated sheet covering content type split, top countries, genre frequency, and rating breakdown
Output: netflix_clean_data.csv exported as UTF-8 CSV for SQL import
The cleaned CSV was imported into a PostgreSQL database (netflix_db). A dedicated setup script handles table schema creation and data import. Six analysis scripts each address a specific analytical theme, producing result-sets exported as CSVs for Power BI.
Database setup (0_db_&_table_creation_and_setup.sql):
CREATE TABLE netflix (
show_id VARCHAR(10) PRIMARY KEY,
type VARCHAR(10),
title VARCHAR(200),
director VARCHAR(300),
cast TEXT,
country VARCHAR(200),
date_added DATE,
release_year INT,
rating VARCHAR(20),
duration VARCHAR(20),
listed_in VARCHAR(200),
description TEXT
);
COPY netflix
FROM 'your/local/path/to/netflix_clean_data.csv'
WITH (FORMAT csv, HEADER true, DELIMITER ',', ENCODING 'UTF8');All 16 result-set CSVs were imported into Power BI via Power Query. Data types were verified and column names standardized before loading. DAX measures were written in a dedicated _Measures table. The dashboard covers a home navigation page and 4 analytical pages.
| # | Problem Statement | Approach |
|---|---|---|
| 1 | What is the overall split between Movies and TV Shows, and what percentage does each hold? | Grouped by type, counted rows, calculated percentage using window-level SUM() OVER() |
| 2 | What is the rating-wise breakdown of all content and what percentage does each rating hold? | Grouped by rating, counted rows per rating, derived percentage via window sum |
| # | Problem Statement | Approach |
|---|---|---|
| 3 | How has Netflix's content count grown year over year, and what is the YoY growth %? | Extracted year from date_added, used two chained CTEs with LAG() to compare each year against the previous |
| 4 | What did Netflix's cumulative library size look like at any point in time? | Used SUM() OVER() with ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW on top of yearly counts CTE |
| — | (Bonus) How does yearly content addition differ between Movies and TV Shows? | Conditional aggregation — SUM(CASE WHEN type = 'Movie' THEN 1 ELSE 0 END) to pivot content types into columns |
| # | Problem Statement | Approach |
|---|---|---|
| 5 | What are the most common individual genres when the multi-genre column is properly split? | Used UNNEST(STRING_TO_ARRAY(listed_in, ',')) to explode comma-separated genres into individual rows, then grouped and counted |
| 6 | Which genre dominated Netflix's catalog for each year? | Three-CTE chain — explode genres, count by year and genre, then DENSE_RANK() OVER(PARTITION BY year) to find rank 1 per year |
| # | Problem Statement | Approach |
|---|---|---|
| 7 | Who are the top 10 most frequent directors on Netflix? | Grouped by director, filtered out Unknown, ordered by count descending with LIMIT 10 |
| 8 | Who are the top 5 directors separately for Movies and TV Shows? | Two CTEs — first counts by director and type, second applies DENSE_RANK() OVER(PARTITION BY type), filtered to rnk <= 5 |
| 9 | Which directors have worked across both Movies and TV Shows? | Grouped by director, used HAVING COUNT(DISTINCT type) = 2 to keep only cross-format directors, STRING_AGG to display both types |
| # | Problem Statement | Approach |
|---|---|---|
| 10 | Which are the top 15 content producing countries? | Used UNNEST(STRING_TO_ARRAY(country, ',')) to handle multi-country rows, grouped and counted individual countries |
| 11 | Which countries produce above average content volume? | Computed per-country counts, used AVG() OVER() as a window function to compare each country against the overall average |
| 12 | What is the Movies vs TV Shows breakdown per top producing country? | Three-CTE chain — expand countries with UNNEST carrying type, conditional aggregation for movie/tv counts, window AVG for above-average status filter |
| # | Problem Statement | Approach |
|---|---|---|
| 13 | What is the overall rating distribution and percentage per rating? | Grouped by rating, counted rows, derived percentage via SUM(COUNT(*)) OVER() |
| 14 | How is content distributed across audience categories and what % does each hold? | Used CASE WHEN to classify 14 ratings into 5 audience bands inside a CTE, then grouped and counted in the outer query |
| 15 | For each audience category, what is the Movies vs TV Shows breakdown? | Extended Problem 14's CTE with conditional aggregation — SUM(CASE WHEN type = 'Movie' THEN 1 ELSE 0 END) alongside the audience classification |
Covers Problems 1, 2, 13, 14, 15 | From: 1_content_distribution.sql & 6_rating_distribution.sql
.png)
Covers Problems 3, 4 and Bonus query | From: 2_growth_trends.sql
.png)
Covers Problems 5, 6, 7, 8, 9 | From: 3_genre_analysis.sql & 4_director_analysis.sql
.png)
Covers Problems 10, 11, 12 | From: 5_country_analysis.sql
.png)
- Netflix is predominantly a Movie platform — 69.62% of its catalog consists of Movies, yet TV Show additions have grown at a faster rate since 2016, signaling a gradual strategic shift toward serialized content
- Content growth was explosive between 2015 and 2019 — Netflix's library scaled from under 500 titles to over 5,000 in just four years; growth has visibly slowed post-2019
- Adult content dominates the catalog — 45.52% of all content falls under the Adult audience category, positioning Netflix as a mature-content platform rather than a family-first service
- International Movies and Dramas are the most prevalent genres — reflecting Netflix's strategy of acquiring broad, globally accessible content for diverse international subscriber bases
- The United States leads content production by a significant margin — but India, the United Kingdom, and Japan are substantial contributors, highlighting Netflix's active localization strategy
- Director talent pools are largely siloed by content type — very few directors have directed both Movies and TV Shows on the platform, suggesting Netflix's Movie and TV production pipelines operate largely independently
Step 1 — Excel Cleaning
- Open
data/Raw data/netflix_titles_raw_data.csvin Excel via Data → Get & Transform → From Text/CSV (UTF-8 encoding) - Apply the cleaning steps documented in Stage 1 above
- Export the cleaned sheet as
netflix_clean_data.csvintodata/Cleaned data/csv/
Step 2 — PostgreSQL Setup & Analysis
- Install PostgreSQL and open pgAdmin 4
- Create a new database named
netflix_db - Open
Scripts/0_db_&_table_creation_and_setup.sqlin VS Code — update the file path in the COPY command to your local path — then execute - Run each analysis script from
Scripts/in numbered order (01 through 06) - After running each script, export the result set as CSV into
result_set_csv_files/
Step 3 — Power BI Dashboard
- Open Power BI Desktop
- Open
Dashboard/netflix_dashboard.pbix - If prompted to update data source paths, redirect to your local
result_set_csv_files/folder - Click Refresh — all visuals will populate
Harsh Menon — Aspiring Data Analyst — Excel · SQL · Power BI
