utils/tools.py

from joblib import dump, load 
import pandas as pd
from sklearn import metrics
from flask import flash
import numpy as np 
import pandas as pd 
from sklearn.preprocessing import LabelEncoder
from sklearn.metrics.pairwise import cosine_similarity
from sklearn import metrics


def data_similarity(df,pt,index,column,value):
    # index fetch
    index = np.where(pt.index==index)[0][0]
    similarity_scores = cosine_similarity(pt)
    similar_items = sorted(list(enumerate(similarity_scores[index])),key=lambda x:x[1],reverse=True)[1:2]
    
    data = []
    for i in similar_items:
        item = []
        temp_df = df[df['index'] == pt.index[i[0]]]
        item.extend(list(temp_df.drop_duplicates(index)[value].values))
        #item.extend(list(temp_df.drop_duplicates(index)[column].values))
        #item.extend(list(temp_df.drop_duplicates(index)[index].values))
        
        data.append(item) 
    list = [item.item() if isinstance(item, np.generic) else item for sublist in data for item in sublist]
    
    original_values = [list['Change_cts_value'].inverse_transform([val]) for val in list]
    
    return original_values  
    
def recommendation_generator(df):
    try:
        pivot_cts = df.pivot_table(index='EngCts', columns='MkblCts', values='Change_cts_value')
        pivot_shp = df.pivot_table(index='EngShp', columns='MkblShp', values='change_shape_value') 
        pivot_qua = df.pivot_table(index='EngQua', columns='MkblQua', values='Change_quality_value')
        pivot_col = df.pivot_table(index='EngCol', columns='MkblCol', values='Change_color_value')
        pivot_cut = df.pivot_table(index='EngCut', columns='MkblCut', values='Change_cut_value')
        
        #==============================================================================
        # # Recommendation
        #==============================================================================
        cts_data = data_similarity(df,pivot_cts,'EngCts','MkblCts','Change_cts_value')
        shp_data = data_similarity(df,pivot_shp,'EngShp','MkblShp','Change_shape_value')
        qua_data = data_similarity(df,pivot_qua,'EngQua','MkblQua','Change_quality_value')
        col_data = data_similarity(df,pivot_col,'EngCol','MkblCol','Change_color_value')
        cut_data = data_similarity(df,pivot_cut,'EngCut','MkblCut','Change_cut_value')
              
        return cts_data,shp_data,qua_data,col_data,cut_data
    
    except Exception as e:
        flash(f'Error generating recommendation: {e}', 'error')
        return None

def classification_report(df):
    try:
        classifcation_data = df[["EngGraphCts","EngCts","EngShp","EngQua","EngCol","EngCut","EngPol","EngSym","EngFlo","EngNts","EngMikly","EngLab","EngAmt",
                                 "MkblCts","MkblShp","MkblQua","MkblCol","MkblCut","MkblPol","MkblSym","MkblFlo","MkblNts","MkblMikly","MkblLab","MkblAmt"]] 
        
        #==============================================================================
        # # Feature Engineering to generate new columns  
        #==============================================================================
        # Make predictions
        classifcation_data["Cts_diff_eng_mkbl"] = round(classifcation_data["EngCts"] - classifcation_data["MkblCts"],2)

        # Create a new column 'Change_Label' based on the values in 'Cts_diff_eng_mkbl'
        classifcation_data['Change_cts_value'] = classifcation_data['Cts_diff_eng_mkbl'].apply(
            lambda x: str(x)+' negative change' if x < 0 else (str(x)+' positive change' if x > 0 else 'no change')
        )
        
        # Create a new column 'Shape_Change' based on the values in 'EngShp' and 'MkblShp'
        classifcation_data['Change_shape_value'] = classifcation_data.apply(
            lambda row: str(row['EngShp'])+' to '+str(row['MkblShp'])+' shape change' if row['EngShp'] != row['MkblShp'] else 'shape not change', axis=1
        )
        
        # Create a new column 'quality_Change' based on the values in 'EngQua' and 'MkblQua'
        classifcation_data['Change_quality_value'] = classifcation_data.apply(
            lambda row: str(row['EngQua'])+' to '+str(row['MkblQua'])+' quality change' if row['EngQua'] != row['MkblQua'] else 'quality not change', axis=1
        )
        
        # Create a new column 'color_Change' based on the values in 'EngCol' and 'MkblCol'
        classifcation_data['Change_color_value'] = classifcation_data.apply(
            lambda row: str(row['EngCol'])+' to '+str(row['MkblCol'])+' color change' if row['EngCol'] != row['MkblCol'] else 'color not change', axis=1
        )
        
        # Create a new column 'cut_Change' based on the values in 'EngCut' and 'MkblCut'
        classifcation_data['Change_cut_value'] = classifcation_data.apply(
            lambda row: str(row['EngCut'])+' to '+str(row['MkblCut'])+' cut change' if row['EngCut'] != row['MkblCut'] else 'cut not change', axis=1
        )
        
        #==============================================================================
        # # Label Encoding and storing the label encoders 
        #==============================================================================
        
        # Get list of categorical variables
        s = (classifcation_data.dtypes =="object")
        object_cols = list(s[s].index)
        print("Categorical variables:")
        print(object_cols)

        # Make copy to avoid changing original data 
        label_data = classifcation_data.copy()

        # Apply label encoder to each column with categorical data
        label_encoder = LabelEncoder()
        for col in object_cols:
            label_data[col] = label_encoder.fit_transform(label_data[col])
            dump(label_encoder, f"./AI_In_Diamond_Industry/Label_encoders/label_encoder_{col}.joblib")   
                
        label_data.head()
        
        #==============================================================================
        # # recommendation_system
        #==============================================================================
        df=classifcation_data.copy()
        
        =recommendation_generator(df)
        
        
        
        
        return label_data
    except Exception as e:
        flash(f'Error generating classification report: {e}', 'error')
        return None