🎬 Market Trends & Genre Success: Analysis of 9,068 Films

Project Overview

What drives content success at scale? This project is a large-scale analysis of the film industry, examining the relationship between production budgets, genre clusters, and market success (box office & audience reception) over a 20-year period (1997–2017).

Using longitudinal modeling and correlation network analysis on 9,068 films, I identified how genres co-occur, evolve, and compete, providing a data-driven framework for understanding content market dynamics.

📄 Publication: Francemone, C. J., Kryston, K., & Grizzard, M. (2023). Examining genre success, co-occurrence, release, and production of 9,068 films over twenty years. Poetics, 99, 1-13.

Read the Full Paper (PDF)

Note: This project uses SPSS for statistical modeling and R for network visualization. The analytical approach is tool-agnostic and transferable to Python-based workflows (e.g., pandas, networkx).

---

Key Findings & Strategic Insights

1. The Shift to "Event" Cinema

The data shows a clear divergence in how studios allocate resources.

• The Finding: Production budgets are heavily skewed toward "Sensation Seeking" genres like Action and Adventure. Conversely, traditional "Classic" genres like Comedy and Romance show negative or non-significant correlations with budget.
• Strategic Insight: Studios are moving away from mid-budget, diversified portfolios. Instead, they are consolidating resources into high-cost, high-reward films that drive mass attendance, while lower-budget genres are likely migrating to streaming or facing reduced production.

2. Release Window Optimization

Studios are not just releasing films; they are timing them strategically across the calendar year.

• The Finding: Release dates are highly segmented. "Sensation Seeking" genres (Action, Sci-Fi) are significantly more likely to release in Summer (capturing leisure time), while "Prestige" genres (Drama, Biography) cluster in Fall/Winter (targeting award consideration).
• Strategic Insight: Distributors optimize release schedules to align with audience availability and industry cycles, where they prioritize entertainment value when users have time off (Summer) and artistic merit during awards season (Winter).

3. Mapping Genre Co-occurence (Cluster Analysis)

Using correlation clustering, I identified that the film market isn't 22 separate genres, but rather it is organized into four distinct ecosystems.

• The Finding: The network analysis revealed four stable clusters: Verisimilitude (Bio/Doc/History), Sensation Seeking (Action/Horror/Sci-Fi), Family Fare (Animation/Fantasy), and Classic (Drama/Romance).
• Strategic Insight: These clusters reveal underlying audience motivations. For example, the "Family Fare" cluster frequently integrates elements from other popular genres (like Action), suggesting that successful family content is often a hybrid designed to capture multiple demographic segments.

---

Methodology

• Data Source: Aggregated dataset of 9,068 films (1997–2017) sourced from OMDb and IMDb.
• Analysis Strategy:
  • Longitudinal Regression: To track genre popularity and production shifts over a 20-year window.
  • Correlation Network Analysis: To map the co-occurrence of genres (e.g., how often Sci-Fi appears with Action).
• Tools:
  • SPSS: Descriptive statistics, ANOVA, and regression modeling.
  • R: Network visualization of genre clusters.

---

Repository Structure

/Market-Trends-and-Genre-Success-of-9-068-Films
│
├── /data
│   ├── Main_Data.csv                          <-- The primary dataset (converted from SAV for accessibility)
│   ├── Genre_Over_Time_Data.csv               <-- Reworked dataset for time-series analysis
│   ├── Main_Data.sav                          <-- Original .sav files (SPSS)
│   └── Genre_Over_Time_Data.sav              
│
├── /analysis
│   ├── Main_Syntax.sps                        <-- Syntax for main analyses
│   └── Genre_Over_Time_Syntax.sps             <-- Syntax for longitudinal analyses
│
├── /visualization
│   ├── Genre_Network.Rmd                      <-- R Markdown generating the co-occurrence network
│   └── Genre_Network_Viz.png                  <-- Visualization of genre clusters
│
└── /results
│   ├── Main_Results.csv                       <-- CSV file containing main output
│   └── Genre_Over_Time_Results.csv            <-- CSV file containing longitudinal output
│
└── Francemone_Genre_Success_2023.pdf          <-- Full published manuscript

---

My Role

Lead Researcher (First Author)

• Aggregated and cleaned 20 years of film data (N = 9,068) from multiple sources.
• Designed the longitudinal analysis framework to track genre trends over time.
• Conducted all statistical modeling (SPSS) and network visualization (R).
• First author on the peer-reviewed publication in Poetics.