This project aims to explore and analyze the characteristics and performance of Netflix content using publicly available data from Kaggle.
Through a series of business-oriented questions, I investigate how different factors — such as content type (Movie vs. Show), release year, age certification — relate to viewer ratings and engagement.
The main goal is to generate actionable insights that could help answer executive-level questions like:
- What type of content is better received by audiences?
- Are certain age groups or genres associated with higher-rated content?
- How do release trends affect viewer satisfaction?
To answer these questions, I perform an extensive exploratory data analysis (EDA) using Python (pandas, seaborn, matplotlib), and visualize key insights with Tableau.
This project is part of my data science portfolio and is designed to showcase my skills in:
- Data cleaning and transformation with
pandas - Exploratory analysis and hypothesis building
- Visual storytelling with
matplotlib,seabornandTableau - Business thinking and insight generation
To investigate this, I analyzed IMDb ratings by content type (MOVIE vs SHOW) and followed a structured questioning process.
I began by calculating the mean IMDb score for each content type:
- Movies: 6.27
- Shows: 7.02
This raised the question:
"Could this difference in average be due to the different number of entries for each content type in the dataset?"
Next, I checked the dataset size:
- Movies: 3,407 entries
- Shows: 1,876 entries
Since there are nearly twice as many movies as shows, this could introduce more variability and possibly more outliers, which may pull the average IMDb score for movies down.
So I asked myself:
"Would the median score — which is more resistant to outliers — still favor shows?"
- Movie median: 6.4
- Show median: 7.2
Even when using median, TV shows outperform movies.
Interpretation:
The result is not driven by a few high-rated shows or low-rated movies - the difference is consistent across the distribution.
To test whether outliers could be affecting this result, I calculated the number and percentage of outliers based on the 1.5 x IQR rule.
- Outliers in movies: 54 (≈1.02%)
- Outliers in shows: 44 (≈0.83%)
New question:
Is this small difference in outlier percentage enough to impact the results?
The answer is likely no - the difference in outlier ratio is too small to account for the substantial gap in scores.
Netflix TV shows consistently achieve higher IMDb scores (both mean and median), with only slightly fewer outliers.
This suggests that shows are not only well-produced, but also more consistently appreciated by viewers than movies on the platform.
To visually support the analysis, I plotted a boxplot comparing the IMDb score distribution of movies and TV shows. The plot helps to see:
- The central tendency (median line)
- The spread of ratings (IQR)
- Outliers (dots)
To understand whether content quality is improving or declining over time, I analyzed the average IMDb scores of Netflix movies and shows by their release year.
- Especially for TV shows, there appears to be significant fluctuation across years, likely due to fewer entries in the dataset. This makes year-over-year trends for shows less stable.
- For both content types, there is no strong or consistent upward or downward pattern. However, the general trend line suggests that IMDb scores may be gradually declining over time.
- This could indicate either a shift in audience expectations, content saturation, or simply a change in rating behavior over the years.
📌 While not conclusive, the data implies that Netflix content — particularly movies — might be receiving slightly lower ratings in more recent years.
I created a dynamic dashboard using Tableau to explore key findings with interactive filters.
🔗 View the live dashboard here:
👉 Netflix Data Dashboard on Tableau Public