andreachello

# THIS SCRIPT USES THE LIBRARY AT:
# https://github.com/hzeller/rpi-rgb-led-matrix
# BE SURE TO CLONE IT AND READ THE README, as highlighted in the video :)


import os, time, threading, random
import feedparser
from PIL import Image, ImageFont, ImageDraw
from random import shuffle

andreachello

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
%matplotlib inline

# Import the data using the file path

data = pd.read_csv('Ames_Housing_Sales.csv')
df = data.copy()

andreachello

# Get a Pd.Series consisting of all the string categoricals
one_hot_encode_cols = data.dtypes[data.dtypes == object]  # filtering by string categoricals
one_hot_encode_cols = one_hot_encode_cols.index.tolist()  # list of categorical fields

# Encode these columns as categoricals so one hot encoding works on split data (if desired)
for col in one_hot_encode_cols:
    data[col] = pd.Categorical(data[col])

# Do the one hot encoding
data = pd.get_dummies(data, columns=one_hot_encode_cols, drop_first=True)

andreachello

from sklearn.linear_model import LinearRegression
from sklearn.preprocessing import StandardScaler

lr = LinearRegression()

y_s_col = "SalePrice"

X_s = data.drop(y_s_col, axis=1)
y_s = data[y_s_col]

andreachello

# Create a list of float colums to check for skewing
mask = df.dtypes == float
float_cols = df.columns[mask]

skew_limit = 0.75 # define a limit above which we will log transform
skew_vals = df[float_cols].skew()

# Showing the skewed columns
skew_cols = (skew_vals
             .sort_values(ascending=False)

andreachello

from sklearn.model_selection import train_test_split
y_col = 'SalePrice'

# Split the data that is one-hot encoded
feature_cols = [x for x in data_ohc.columns if x != y_col]
X_data_ohc = data_ohc[feature_cols]
y_data_ohc = data_ohc[y_col]

X_train_ohc, X_test_ohc, y_train_ohc, y_test_ohc = train_test_split(X_data_ohc, y_data_ohc, 
                                                    test_size=0.3, random_state=42)

andreachello

from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_squared_error

LR = LinearRegression()

# Data that have been one-hot encoded
LR = LR.fit(X_train_ohc, y_train_ohc)
y_train_ohc_pred = LR.predict(X_train_ohc)
y_test_ohc_pred = LR.predict(X_test_ohc)

andreachello

plt.figure(figsize=(12,5))

sns.set_context('talk')
sns.set_style('ticks')
sns.set_palette('dark')

ax = plt.axes()
# we are going to use y_test_ohc, y_test_ohc_pred
ax.scatter(y_test_ohc, y_test_ohc_pred, alpha=.5)

andreachello

from sklearn.preprocessing import StandardScaler, PolynomialFeatures
from sklearn.model_selection import KFold, cross_val_predict
from sklearn.linear_model import LinearRegression, Lasso, Ridge
from sklearn.metrics import r2_score
from sklearn.pipeline import Pipeline

kf = KFold(shuffle=True, random_state=72018, n_splits=3)

X = data.drop('SalePrice', axis=1)
y = data.SalePrice

andreachello

from sklearn.linear_model import RidgeCV
from sklearn.metrics import mean_squared_error

# root-mean-squared error function
def rmse(ytrue, ypredicted):
    return np.sqrt(mean_squared_error(ytrue, ypredicted))

alphas = [0.005, 0.05, 0.1, 0.3, 1, 3, 5, 10, 15, 30, 80]

ridgeCV = RidgeCV(alphas=alphas,