This is an app that generates comps for real estate properties. It uses euclidean distances between physical coordinates and property feature strings to calculate the most similar properties.

Instructions:

pip install fuzzywuzzy geopy thefuzz

Step 1:

load the JSON file with open('appraisals_dataset.json', 'r') as f: data = json.load(f) df = pd.DataFrame(data) df = pd.json_normalize(df['appraisals'])

Step 2:

Clean the features in dataframe in the section: Data Cleaning Pipeline Enter which row in the variable (i) for your training data

Step 3:

Format the data for Training in the Training secion

Step 4:

Use Euclidean Distances on the data points. Validate with RandomForestRegressor