Spaces:
Build error
Build error
File size: 9,353 Bytes
ad98fbf e2eacae ad98fbf |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 |
# Contains classes and functions for handling and transforming
# features based on the JSON file information.
import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler, MinMaxScaler, LabelEncoder, OrdinalEncoder, OneHotEncoder
import streamlit as st
class FeatureHandler:
def __init__(self, json_content):
self.json_content = json_content
def impute_missing_values(self, feature_details, X_train, X_test=None):
mean_impute_features = []
median_impute_features = []
mode_impute_features = []
custom_impute_features = []
for feature in X_train.columns:
details = feature_details[feature]['feature_details']
if details["missing_values"]:
if details["impute_with"].lower() == "mean":
mean_impute_features.append(feature)
elif details["impute_with"].lower() == "median":
median_impute_features.append(feature)
elif details["impute_with"].lower() == "mode":
mode_impute_features.append(feature)
elif details["impute_with"].lower() == "custom":
custom_impute_features.append(feature)
if mean_impute_features:
X_train[mean_impute_features] = X_train[mean_impute_features].fillna(X_train[mean_impute_features].mean())
if median_impute_features:
X_train[median_impute_features] = X_train[median_impute_features].fillna(X_train[median_impute_features].median())
if mode_impute_features:
X_train[mode_impute_features] = X_train[mode_impute_features].fillna(X_train[mode_impute_features].mode().iloc[0])
if custom_impute_features:
for feature in custom_impute_features:
X_train[feature] = X_train[feature].fillna(feature_details[feature]['feature_details']['custom_impute_value'])
if X_test is not None:
if mean_impute_features:
X_test[mean_impute_features] = X_test[mean_impute_features].fillna(X_train[mean_impute_features].mean())
if median_impute_features:
X_test[median_impute_features] = X_test[median_impute_features].fillna(X_train[median_impute_features].median())
if mode_impute_features:
X_test[mode_impute_features] = X_test[mode_impute_features].fillna(X_train[mode_impute_features].mode().iloc[0])
if custom_impute_features:
for feature in custom_impute_features:
X_test[feature] = X_test[feature].fillna(feature_details[feature]['feature_details']['custom_impute_value'])
return X_train, X_test
# TODO: Add imputation for categorical features
def scale_features(self, feature_details, X_train, X_test=None):
min_max_scaler_features = []
standard_scaler_features = []
# for feature, details in feature_details.items():
for feature in feature_details.keys():
details = feature_details[feature]['feature_details']
if details.get("rescaling"):
if details["rescaling"]!= "No rescaling" and details["scaling_type"] == "MinMaxScaler" :
min_max_scaler_features.append(feature)
elif details["rescaling"] != "No rescaling" and details["scaling_type"] == "StandardScaler" :
standard_scaler_features.append(feature)
if min_max_scaler_features:
scaler = MinMaxScaler()
X_train_scaled = scaler.fit_transform(X_train[min_max_scaler_features])
X_train[min_max_scaler_features] = X_train_scaled
if standard_scaler_features:
scaler = StandardScaler()
X_train_scaled = scaler.fit_transform(X_train[standard_scaler_features])
X_train[standard_scaler_features] = X_train_scaled
if X_test is not None:
if min_max_scaler_features:
X_test_scaled = scaler.fit_transform(X_test[min_max_scaler_features])
X_test[min_max_scaler_features] = X_test_scaled
if standard_scaler_features:
X_test_scaled = scaler.fit_transform(X_test[standard_scaler_features])
X_test[standard_scaler_features] = X_test_scaled
return X_train, X_test
def encode_features(self, feature_details, X_train, X_test=None):
ordinal_encoder_cols = []
one_hot_encoder_cols = []
for feature in X_train.columns:
if feature_details[feature]["feature_variable_type"] == "object":
details = feature_details[feature]['feature_details']
if details["encoding"] == "OridnalEncoder":
ordinal_encoder_cols.append(feature)
elif details["encoding"] == "OneHotEncoder":
one_hot_encoder_cols.append(feature)
if ordinal_encoder_cols:
ordinal_encoder = OrdinalEncoder()
X_train[ordinal_encoder_cols] = ordinal_encoder.fit_transform(X_train[ordinal_encoder_cols])
if X_test is not None:
X_test[ordinal_encoder_cols] = ordinal_encoder.transform(X_test[ordinal_encoder_cols])
if one_hot_encoder_cols:
one_hot_encoder = OneHotEncoder( drop="first", sparse_output=False)
temp_df = pd.DataFrame(one_hot_encoder.fit_transform(X_train[one_hot_encoder_cols]),
columns=one_hot_encoder.get_feature_names_out(),
index=X_train.index)
X_train = X_train.drop(one_hot_encoder_cols, axis=1)
X_train = pd.concat([X_train, temp_df], axis=1)
if X_test is not None:
temp_df = pd.DataFrame(one_hot_encoder.transform(X_test[one_hot_encoder_cols]),
columns=one_hot_encoder.get_feature_names_out(),
index=X_test.index)
X_test = X_test.drop(one_hot_encoder_cols, axis=1)
X_test = pd.concat([X_test, temp_df], axis=1)
return X_train, X_test
def transform_X_features(self, X_train, X_test, feature_details):
X_train_transformed, X_test_transformed = self.impute_missing_values(feature_details, X_train, X_test)
X_train_transformed, X_test_transformed = self.encode_features(feature_details, X_train_transformed, X_test_transformed)
X_train_transformed, X_test_transformed = self.scale_features(feature_details, X_train_transformed, X_test_transformed)
return X_train_transformed, X_test_transformed
# tokenize and hash the target variable
def tokenize_target_variable(self, y_train, y_test):
details = self.json_content["design_state_data"]["feature_handling"]
feature_details = details[y_train.name]["feature_details" ]
if feature_details["text_handling"] == "Tokenize and hash":
# tokenize the target variable
label_encoder = LabelEncoder()
y_train_tokenized = y_train.apply(lambda x: x.split("-")[1])
y_train_encoded = label_encoder.fit_transform(y_train_tokenized)
y_test_tokenized = y_test.apply(lambda x: x.split("-")[1])
y_test_encoded = label_encoder.transform(y_test_tokenized)
return y_train_encoded, y_test_encoded
def label_encode_target_variable(self, y_train, y_test):
label_encoder = LabelEncoder()
y_train_encoded = label_encoder.fit_transform(y_train)
y_test_encoded = label_encoder.transform(y_test)
return y_train_encoded, y_test_encoded
def transform_y_features(self, y_train, y_test, feature_details, target_variable):
if feature_details[target_variable]["feature_variable_type"] == "object":
if feature_details[target_variable]["feature_details"]["text_handling"] == "Tokenize and hash":
y_train_transformed, y_test_transformed = self.tokenize_target_variable(y_train, y_test)
elif feature_details[target_variable]["feature_details"]["text_handling"] == "Label Encoding":
y_train_transformed, y_test_transformed = self.label_encode_target_variable(y_train, y_test)
return y_train_transformed, y_test_transformed
else:
return y_train, y_test
def get_split_dataset(self, selected_features):
design_state = self.json_content["design_state_data"]
dataset = design_state["session_info"]["dataset"]
target_variable = design_state["target"]["target"]
train_info = design_state["train"]
train_ratio = train_info["train_ratio"]
random_seed = train_info["random_seed"]
DATASET_PATH = "data/"+dataset
df = pd.read_csv(DATASET_PATH)
X = df[selected_features]
Y = df[target_variable]
X_train, X_test, y_train, y_test = train_test_split(X, Y, train_size=train_ratio,
random_state=random_seed)
return X_train, X_test, y_train, y_test |