linear regression to predict solpunk price based on sales data

solpunks.py

import json
import pandas as pd
from sklearn import linear_model
from sklearn.model_selection import train_test_split

def list_traits(a):
    return list(map(lambda x: " ".join(x.split(' ')[0:-1]), a.split(',')))


# Load list of all punks
with open('solpunks.json', 'r') as pr:
    data = pr.read()
    punk_json = json.loads(data)

punks = {}
all_traits = set()
for punk in punk_json['punks']:

    # Skip SatoshiStreetBets
    if 'id' not in punk:
        continue
    if punk['attributes'] == 'NULL':
        punk_traits = ['NoTraits']
    else:
        punk_traits = list_traits(punk['attributes']) 
    punks[punk['id']] = {
        "ranking": punk["ranking"],
        "skin": punk["skin"],
        "type": punk["type"],
        "traits": punk_traits
    }

    for trait in punk_traits:
        all_traits.add(trait)

for punk_id,punk in punks.items():

    my_traits = punk['traits']

    del punk['traits']
    for trait in all_traits:
        punk[trait] = False
    for trait in my_traits:
        punk[trait] = True



# Load pricing data
with open('all_solpunks_sold.json') as sales_data:
    sales = json.loads(sales_data.read())
    sales.reverse()

for sale in sales:
    punk_id = sale['name']
    # Skip SatoshiStreetBets
    try:
        punk_id = punk_id.split('#')[1].zfill(4)
    except:
        continue
    punks[punk_id]['price'] = sale['price']



df = pd.DataFrame(punks.values())

non_dummy_cols = ['ranking', 'price']
X = df
dummy_cols = list(set(X.columns) - set(non_dummy_cols))
all_X = pd.get_dummies(X, columns=dummy_cols, drop_first=True)
X = all_X[all_X.price.notnull()]
Y = X['price']

# We don't want to train on price, we want to predict it
del X['price']


X_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size = .20, random_state = 40)
regr = linear_model.LinearRegression()
regr.fit(X_train, Y_train)
predicted = regr.predict(X_test)

import pdb
pdb.set_trace()

# To get prediction for punk id #9203, use
# regr.predict(all_X)[9203]