Project Overview:

Our project has two primary goals: identify and predict trends in NYC building permits and classify pre-1940 buildings based on recent maintenance. These efforts provide critical insights for our investment firm stakeholders, helping them seize valuable real estate opportunities. The Juptyer notebook with code to this project can be found here: Notebook

Business Understanding:

Real estate, a key economic driver in NYC, continually evolves due to the constant demand for new construction. By understanding these trends, stakeholders, like investment firms, make informed decisions and capitalize on lucrative opportunities.

Stakeholder: MetroStar Real Estate Investment Firm

MetroStar Real Estate Investment Firm is a NYC-based entity keen on leveraging data to understand and predict city construction trends. By focusing on new developments and buildings ripe for renovation, the firm aims to capitalize on investment potential. Our data analysis and predictive modeling services will offer crucial insights, shaping their future investment strategies effectively.

Data Understanding:

We used two datasets, DOB NOW: Build – Approved Permits and PLUTO, which offer comprehensive information on building permits and extensive land use details respectively.

Old Buildings Classification & Maintenance

We identified pre-1940 NYC buildings and categorized them based on maintenance status post-2018:

• Maintained Buildings: Received upkeep since 2018, potentially in better condition but could still benefit from improvements.
• Unmaintained Buildings: Lacked maintenance since 2018, offering significant renovation and value enhancement opportunities.

Classification Model Insights The 'Approved Year' feature played a critical role in our model due to its strong connection with recent renovation activity. This aligns with our goal of identifying older buildings in top community boards that haven't seen recent upgrades, enabling targeted renovation planning and effective resource allocation.

However, the tight correlation between 'Approved Year' and the target variable 'recent_permit' might lead some to view this as data leakage. Hence, while the model successfully predicts recent renovations, its predictions may not go beyond what can be directly inferred from the 'Approved Year'.

In attempt to adress the data leakage and improve our model's performance, we identified and removed features potentially causing data leakage, specifically "Approved Date", "Approved Year", and "Approved Month". Despite removing these features, our model achieved a remarkable 99% accuracy. The "C B NO" feature was retained, given its significance in prediction. Our efforts confirm that our model can deliver reliable predictions even with these changes, demonstrating its robustness.

Data Cleaning and Preprocessing:

We cleaned and preprocessed the datasets by handling missing values, converting data types, and extracting useful features.

Exploratory Data Analysis (EDA):

We conducted an EDA to understand data trends, distributions, and relationships, and identified top community boards within each borough based on building permit growth.

Here we can see the distribution of permits approved by borough:

Time Series Analysis and Forecasting:

We performed time series analysis on the permit data and utilized several models (ARIMA, SARIMA, Prophet, LSTM) to forecast permit approval trends over the next year for each top community board.

Prediction results for community board with the most growth in the BRONX.

• RMSE: 13.68
  • MAE: 10.31

Our LSTM model's performance in predicting the number of building permits for the top community board in the Bronx is promising. The model achieved a Root Mean Squared Error (RMSE) of 13.68 and a Mean Absolute Error (MAE) of 10.31. These metrics suggest that our forecasts are close to the actual values, demonstrating the model's effectiveness in predicting building permit trends for future strategic planning.

---

Prediction Results:

Our LSTM model for predicting community board growth across the boroughs provided the following results:

• Bronx (CB 204.0):

  • RMSE: 13.68
  • MAE: 10.31

• Brooklyn (CB 301.0):

  • RMSE: 33.68
  • MAE: 28.21

• Manhattan (CB 105.0):

  • RMSE: 37.96
  • MAE: 30.17

• Queens (CB 407.0):

  • RMSE: 24.04
  • MAE: 18.08

• Staten Island (CB 501.0):

  • RMSE: 21.03
  • MAE: 18.33

The average RMSE across all boroughs for the LSTM model was 26.08. These results show the root mean square error (RMSE) and the mean absolute error (MAE) for each borough's predicted growth. These metrics provide a measure of how well our model fits the observed data - lower values indicate better fit. Remember, these results are meant to serve as a guide and additional comprehensive analysis should always be conducted.

---

---

Location Recommendations:

This analysis allowed us to identify top-performing community boards within each NYC borough, providing valuable insights for potential real estate investments:

1. Brooklyn (CB 301.0 - Neighborhoods like Williamsburg): Brooklyn is known for its diverse culture and creative energy. Its rapid development and top building types such as walk-up apartments and two-family houses signal a thriving real estate market. Commercial garages also feature prominently, underscoring the need for commercial real estate infrastructure in this bustling borough.

2. Bronx (CB 204.0 - Neighborhoods like Highbridge, Concourse): The Bronx offers affordable real estate options and excellent connectivity. It is an emerging market with untapped potential. With a high prevalence of apartment buildings, both with and without stores, there's strong potential for residential and mixed-use buildings. The commonality of two-family brick buildings and one-story retail buildings enhances the appeal of this dynamic borough.

3. Manhattan (CB 105.0 - Neighborhoods like Times Square, Theater District): At the heart of NYC, Manhattan provides prestigious real estate opportunities with potential for attractive returns. The popularity of office buildings and residential buildings with commercial units suggests a high demand for mixed-use properties. The presence of big-box stores indicates lucrative opportunities in the retail sector.

4. Queens (CB 407.0 - Neighborhoods like Flushing, Whitestone): Renowned for its cultural diversity and dynamic economy, Queens provides an opportunity to cater to varied demographics in a growing market. The borough is dominated by one-family houses, both detached and semi-detached, and two-family semi-attached houses, making it ideal for family-oriented residential developments. Brick buildings and garages or gas stations also feature, offering diverse investment opportunities.

5. Staten Island (CB 501.0 - Neighborhoods like St. George, Tompkinsville): Offering suburban charm and scenic landscapes, Staten Island is perfect for family-oriented investments, including opportunities for waterfront properties. Detached one-family and two-family houses are popular, reinforcing the borough's appeal for families. The presence of one-story garages or gas stations and shopping centers indicates a need for commercial infrastructure.

Remember, these community boards can serve as focal points to explore surrounding neighborhoods for further opportunities. Thorough market research and consideration of individual investment goals and risk tolerance are essential before making any investment decisions.

---

Limitations:

While our analysis offers valuable insights, it's important to consider certain limitations:

1. Reliance on Historical Data: Past trends may not guarantee future outcomes due to unpredictable factors such as economic downturns or policy changes.
2. Data Scope: Our focus on building classifications and permits omits other crucial factors like property prices, market demand, and local amenities.
3. Geographical Limitation: Our NYC-centric analysis might not translate to other real estate markets with unique factors.
4. Model Assumptions: Our models assume a steady rate of growth and similar market conditions, which might not hold true.

Further Analysis:

Our work can be extended in several ways:

1. Incorporate Additional Factors: Enhance analysis accuracy by including elements like real estate prices, neighborhood crime rates, and school district ratings.
2. Detailed Borough Analysis: Understand micro-market trends within individual boroughs for hidden investment opportunities.
3. Refine Predictive Models: Fine-tune models with more complex machine learning algorithms and granular data for accurate future trend predictions.
4. Expand to Other Markets: Apply our methodology to different geographic areas for insights into diverse real estate markets. Certainly, here is a section to help navigate your repository:

Navigating The Repository:

1. .gitignore: This file is used to exclude certain files from the repository. If there are files or paths that we want Git to ignore, we can add rules for them in this file.

2. LICENSE: This file contains information about the license that governs the use and distribution of the repository's content.

3. Notebook.ipynb: This Jupyter Notebook for detailed analysis and findings. contains all the code used for data preprocessing, exploratory data analysis, model training, model evaluation, and other computations.

4. Presentation.pdf: Provides the presentation associated with this project.

5. README.md: This Markdown file provides an overview of the project, describes the business context, data, approach, results, and limitations, and explains how to navigate the repository. If you are viewing the project on GitHub, this is the first file you see on the main page of the repository.

Remember, you can click on any file in the repository to view its contents. If you have any questions or need additional help navigating the repository, don't hesitate to ask.