/machine_learning

Machine learning and data science projects

Primary LanguageJupyter Notebook

Projects Using Machine Learning & Data Science Skills

UnemploymentRate_TimeSeriesPredict.ipynb

US Unemployment Rate Predictions to Track Pandemic Impact in 2020

Objectives:

  • Use time series prediction models to estimate expected US unemployment rate in 2020 in order to accurately measure the effects of the COVID-19 pandemic on US unemployment rate in 2020.
  • Data from 2019 will be used to test time series prediction models and then the best fitting model will be used to forecast expected unemployment rates for 2020 and compared to the actual unemployment figures by different group demographics.

Skills:

  • Data Cleaning
  • Exploratory Data Analysis
  • Feature Engineering
  • Data Visualization
  • Time Series Decomposition
  • Testing Stationarity
  • Training Time Series Models
    • Simple Average
    • Single Exponential
    • Double Exponential
    • Triple Exponential
  • Time Series Forecasting
  • Cross Validation

HousingInsecurity_EvictionRateDeepLearn.ipynb

Housing Insecurity Project: Predicting High Eviction Rates with Deep Learning

Objectives:

  • Use data on housing loss to find factors that can reduce housing insecurity.
  • Explore factors affecting eviction rates and use deep learning to predict areas with higher than average eviction rates.
  • Model can be used to easily identify areas that may have higher than average eviction rates so additional resources can be allocated to reduce housing loss.

Skills:

  • Data Cleaning
  • Exploratory Data Analysis
  • Feature Engineering
  • Testing of Feature Importance
  • Predictive Modeling
  • Hyperparameter Tuning
  • Supervised Models
    • Random Forest Classifier
    • Linear Regression
  • Unsupervised Methods
    • K-Means Clustering
  • Deep Learning Models
    • Constructing Neural Nets
    • Testing Activation Functions
  • Metrics
    • ROC-AUC Curves
    • Loss Functions
    • Accuracy
  • Cross Validation

HousingInsecurity_EvictionRateClustering.ipynb

Housing Insecurity Project: Exploring Factors Causing High Eviction Rates

Objectives:

  • Use data on housing loss to find factors that can reduce housing insecurity.
  • Explore factors affecting eviction rates using unsupervised machine learning (ML) clustering and dimensionality reduction methods.
  • Use results to see if predicting areas with higher than average eviction rates with supervised ML are improved with initial exploration using unsupervised ML.
  • Model can be used to easily identify areas that may have higher than average eviction rates so additional resources can be allocated to reduce housing loss.

Skills:

  • Data Cleaning
  • Exploratory Data Analysis
  • Feature Engineering
  • Testing of Feature Importance
  • Predictive Modeling
  • Hyperparameter Tuning
  • Unsupervised Methods
    • K-Means Clustering
    • Agglomerative Clustering
    • Principal Component Analysis
  • Supervised Models
    • Random Forest Classifier
    • Linear Regression
    • Logistic Regression
  • Cross Validation

HousingInsecurity_EvictionRateModel.ipynb

Housing Insecurity Project: Predicting Areas with High Eviction Rates

Objectives:

  • Use data on housing loss to find factors that can reduce housing insecurity.
  • Predict areas with above average eviction rates based on correlating demographic factors.
  • Model can be used to easily identify areas that may have higher than average eviction rates so additional resources can be allocated to reduce housing loss.

Skills:

  • Data Cleaning
  • Exploratory Data Analysis
  • Feature Engineering
  • Testing of Feature Importance
  • Predictive Modeling
  • Hyperparameter Tuning
  • Training & Testing Models
    • Linear Regression
    • Logistic Regression
    • K-Nearest Neigbors
    • Decision Tree
  • Cross Validation

HousingInsecurity_RentPredictionModel.ipynb

Housing Insecurity Project: Predicting Fair Rental Prices for Social Housing

Objectives:

  • Use data on housing loss to find factors that can reduce housing insecurity.
  • Predict a fair market rent price for social housing based on local housing market information from census data.
  • Model can be used to easily identify overpriced social housing so it can be adjusted accordingly to help reduce housing loss.

Skills:

  • Data Cleaning
  • Exploratory Data Analysis
  • Feature Engineering
  • Testing of Feature Importance
  • Predictive Modeling
  • Hyperparameter Tuning (with GridSearchCV)
  • Training & Testing Models
    • Linear Regression
    • Linear Regression with Polynomial Features
    • Ridge Regression
  • Cross Validation

HousingInsecurity_DataKind_EDA.ipynb

Objectives:

  • Use data on housing loss to find factors that can reduce housing insecurity.
  • Explore whether Florida reserves more units for certain protected groups compared to other locations.

Skills:

  • Exploratory Data Analysis
  • Data Cleaning
  • Key Findings & Insights

FastFoodUSA_EDA_HTest.ipynb

Objectives:

  • Track trends of fast food restaurants in the USA per city/capita.
  • Investigate whether there is a higher number of fast food restaurants per capita in the geographic south of mainland USA.

Skills:

  • Exploratory Data Analysis
  • Data Cleaning
  • Key Findings & Insights
  • Hypothesis Testing