eda

Exploratory Data Analysis (EDA)：

• Data Inventory: Successfully ingested raw dataset containing 10,866 rows and 21 observations.

• Visual Verification: Initial profiling via Table View no significant inconsistencies, spelling or formatting errors detected.

• Missing values :

  • High missing rate: homepage (73% ), tagline (26% ) and keywords (13%) identified, do no directly related to the object, removed.

  • Minor missing rate: genres, product_company and cast, requiring imputation.

• Outlier : Box Plot reveals significant outliers in revenue and budget.

Financial Correlation and Distribution Analysis

• Variable Isolation: Refined the dataset to focus exclusively on core financial metrics (budget, revenue) and their inflation-adjusted counterparts to ensure historical comparability.

• Visual Insights:

  • Correlation: Confirmed a positive trend between budget and revenue.

  • Distribution: Validated a heavily right-skewed via Histogram; confirmed that the vast majority of films earn under $500e.

• Final Output: Exported clean dataset for ROI analysis.

Genre Prevalence Analysis

• Isolation: Used Column Filter to extract only id and genres.

• Normalization: Applied Cell Splitter followed by Ungroup to transform multi-valued genre strings into a list, ensuring each movie is correctly indexed under every applicable category.

• Aggregation: Utilized Group By on the split genre results with a Count aggregation on id to determine the total volume of films produced per category.

• Visualization: The Bar Chart reveals that Drama is the most frequent genre in the dataset ( 1,500 ), followed closely by Comedy and Thriller/Action.

Rating Analysis

• Isolation: Used a Column Filter to isolate popularity, vote_average, and vote_count for analysis.

• Correlation :

  • The Scatter Plot reveals a weak linear correlation between popularity and vote_average.

  • Observation: High-rated films (score > 8) often maintain low popularity scores, suggesting that critical acclaim does not always drive mass-market viral success.

• Outlier and Distribution :Box Plot, Identified significant outliers in vote_count, indicating a small subset of "blockbuster" films receive disproportionately higher audience engagement.

• Final Output: Exported the dataset via CSV Writer for final reporting.

Data cleaning：

• Integrity Enforcement: Applied Duplicate Row Filter to remove redundant records and ensure the uniqueness of each movie .

• Text Normalization:

  • Utilized String Cleaner to strip unnecessary space.

  • Used String Manipulation (Multi Column) to capitalize categories and maintain consistency across text-based fields.

• Temporal Formatting: Converted the release_date column from string format to a standardized Date&Time format.

• Dimensionality Reduction: Implemented a Column Filter to remove low-relevance features such as imdb_id, homepage, and overview to focus on analytical objects.

• Constraint Filtering: Applied a Row Filter to exclude records with unrealistic financial data, specifically targeting movies with budget and revenue below $10,000.

• Missing values: Applied Missing Value node, using median imputation for runtime and mean imputation for vote_average to maintain statistical distribution.

• Correlation Verification: Validated the cleaned dataset through Linear Correlation and Heatmap nodes, confirming strong relationships between revenue, budget, and vote.