README.md

Cervical-Cancer_Diagnosis_Prediction

Variables Used

Load Dataset

Variable Name	Definition
df	whole dataset
X	features
y	target

Split Dataset

Variable Name	Definition
X_train	training set - features
X_vald	validation set - features
X_test	test set - features
y_train	training set - target
y_vald	validation set - target
y_test	test set - target

Data Preprocessing

Missing Data

Variable Name	Definition
X_train_nan	X_train, replaced ‘?’ with nan
X_train_imputed	X_train_nan, nan replaced with column mean
X_vald_nan	X_vald, replaced ‘?’ with nan
X_vald_imputed	X_vald_nan, nan replaced with column mean
X_test_nan	X_test, replaced ‘?’ with nan
X_test_imputed	X_test_nan, nan replaced with column mean

Outliers

Variable Name	Definition
column_train	column index which contains outliers
X_train_outlier	copy of X_train_imputed
X_train_sqrt	X_train_outlier, columns with outlier transformed with square root transformation
column_vald	column index which contains outliers
X_vald_outlier	copy of X_vald_imputed
X_vald_sqrt	X_vald_outlier, columns with outlier transformed with square root transformation
column_test	column index which contains outliers
X_test_noutlier	copy of X_test_imputed
X_test_sqrt	X_test_outlier, columns with outlier transformed with square root transformation

Imbalanced Data

Variable Name	Definition
y_train_reset_index	y_train, index reset
train	concatenation of X_train_sqrt and y_train_reset_index
train_major	train, rows with column ‘Biopsy’ = 0 (majority class)
train_minor	train, rows with column ‘Biopsy’ = 1 (minority class)
train_minor_upsampled	train_minor, resampled to 722 rows
train_upsampled	concatenation of train_minor_upsampled and train_major
X_train_balanced	X_train_sqrt, balanced data
y_train_balanced	y_train, balanced data
y_vald_reset_index	y_vald, index reset
vald	concatenation of X_vald_sqrt and y_vald_reset_index
vald_major	vald, rows with column ‘Biopsy’ = 0 (majority class)
vald_minor	test, rows with column ‘Biopsy’ = 1 (minority class)
vald_minor_upsampled	vald_minor, resampled to 81 rows
vald_upsampled	concatenation of vald_minor_upsampled and vald_major
X_vald_balanced	X_vald_sqrt, balanced data
y_vald_balanced	y_vald, balanced data
y_vald_balanced	y_vald, balanced data
y_test_reset_index	y_test, index reset
test	concatenation of X_test_sqrt and y_test_reset_index
test_major	test, rows with column ‘Biopsy’ = 0 (majority class)
test_minor	test, rows with column ‘Biopsy’ = 1 (minority class)
test_minor_upsampled	test_minor, resampled to 81 rows
test_upsampled	concatenation of test_minor_upsampled and test_major
X_test_balanced	X_test_sqrt, balanced data
y_test_balanced	y_test, balanced data

Data Normalization

Variable Name	Definition
columns_need_norm	Columns in the dataframe that needs to be normalized
X_train_norm	X_train_sqrt, normalized
X_vald_norm	X_vald_sqrt, normalized
X_test_norm	X_test_sqrt, normalized
X_train_norm_balanced	X_train_balanced, normalized
X_vald_norm_balanced	X_vald_balanced, normalized
X_test_norm_balanced	X_test_balanced, normalized

Data Standardization

Variable Name	Definition
columns_need_std	Columns in the dataframe that needs to be standardized
X_train_std	X_train_sqrt, standardized
X_vald_std	X_vald_sqrt, standardized
X_test_std	X_test_sqrt, standardized
X_train_std_balanced	X_train_norm_balanced, standardized
X_vald_std_balanced	X_vald_norm_balanced, standardized
X_test_std_balanced	X_test_norm_balanced, standardized

Feature Selection

Correlation Heatmap

Variable Name	Definition
train_corr	concatenation of X_train_std and y_train_rest_index, to be used in producing correlation matrix
corr_matrix	correlation matrix created with train_corr

Identifying columns

Variable Name	Definition
corr_biopsy	corr_matrix with only the column ‘Biopsy’
drop_features	list of names of columns to be dropped
drop_features_index	list of index of columns to be dropped

Drop low correlation features

Variable Name	Definition
X_train	X_train_std, with less correlatied features removed
X_vald	X_vald_std, with less correlatied features removed
X_test	X_test_std, with less correlatied features removed
X_train_balanced	X_train_std_balanced, with less correlatied features removed
X_vald_balanced	X_vald_std_balanced, with less correlatied features removed
X_test_balanced	X_test_std_balanced, with less correlatied features removed