Loading Evaluation/train_evaluation.py 0 → 100644 +168 −0 Original line number Diff line number Diff line #!/usr/bin/env python3 # -*- coding: utf-8 -*- """ Created on Fri Mar 26 16:11:25 2021 @author: erkinmenderes """ import pandas as pd #GS_annotations file should be in the same directory, else insert filepath output = pd.read_csv("knowlege_path_subset_train.csv", sep=';', error_bad_lines=False) index_list = output.index.tolist() #output["Essay"] += 1 #print(index_list) essay_output_list = output["Essay"].tolist() essay_output_list = essay_output_list[:94] path_list = output["Path"].tolist() gold_df = pd.read_csv("input_subset_train.csv", sep=';', error_bad_lines=False) gold_essay_list = gold_df["Essay"].tolist() reiss_gold = gold_df["Reiss"].tolist() def replacer(list_string): """ Function to replace all the unnecessary characters in the paths in order to further process the data and return clean strings Parameters ---------- list_string : list a list containing strings of words (here: conceptnet paths as strings) Returns ------- str strings cleaned from unwanted and unnecessary characters and tokens """ text = list_string.replace("[", "").replace("]", "").replace('\'', "").replace("\"", "").replace(",", "") return text.split() # create maslow and reiss human needs maslow_human_needs = ["physiological needs", "stability", "love/belonging", "esteem", "spiritual growth"] reiss_motives = ["food", "rest", "health", "save_money", "order", "safety", "romance", "belonging", "family", "contact", "competition", "honor", "approval", "status", "power", "curiosity", "serenity", "idealism", "independent"] # create a cleaned list of paths cleaned_paths = [replacer(path) for path in path_list] def assign_reiss(path_list): """ Assigns Reiss motive to a a graph consisting of a list of its paths Parameters ---------- path_list : list The entire list of subraphs which consist of their paths paths are split into their single units as strings Returns ------- None. """ temp_list = [] human_needs = [] for path in path_list: for word in path: if word in reiss_motives: temp_list.append(word) human_needs.append(temp_list[0]) temp_list = [] return human_needs #assign reiss human needs for every graph via its top ranked path reiss_needs = assign_reiss(cleaned_paths) #maslow needs list to assign maslow need accordingly physiological_needs = ['food', 'rest'] safety = ['health', 'save_money', 'order', 'safety'] love_belonging = ['love', 'belonging', 'family', 'contact'] esteem = ['competition','honor', 'approval', 'status', 'power'] spiritual_growth = ['curiosity', 'serenity','idealism', 'independent'] def assign_maslow(reiss_list): """ Function that assigns corresponding maslow human need given its reiss human need Parameters ---------- reiss_list : list list containing assigned reiss human need for every essay (via the top ranked graphpath) Returns ------- maslow_needs : list list containing corresponding maslow human needs """ maslow_needs = [] for r in reiss_needs: if r in physiological_needs: maslow_needs.append(maslow_human_needs[0]) elif r in safety: maslow_needs.append(maslow_human_needs[1]) elif r in love_belonging: maslow_needs.append(maslow_human_needs[2]) elif r in esteem: maslow_needs.append(maslow_human_needs[3]) else: maslow_needs.append(maslow_human_needs[4]) return maslow_needs maslow_needs = assign_maslow(reiss_needs) # create joint list of reiss and maslow hn_list_full = list(zip(maslow_needs, reiss_needs)) #post-processing for evaluation in reiss reiss_needs = [w.replace("independent", "independence") for w in reiss_needs] reiss_needs = [w.replace("save_money", "savings") for w in reiss_needs] #post-processing for evaluation in maslow maslow_needs = [w.replace("love / belonging", "love/belonging") for w in maslow_needs] # add columns accordingly output["Maslow_predict"] = maslow_needs output["Reiss_predict"] = reiss_needs from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score from sklearn.metrics import ConfusionMatrixDisplay, classification_report, confusion_matrix train_acc = accuracy_score(reiss_gold, reiss_needs) print("Training accuracy: ", train_acc) train_prec = precision_score(reiss_gold, reiss_needs, average='macro') print("Training precision: ", train_prec) train_recall = recall_score(reiss_gold, reiss_needs, average='macro') print("Training recall: ", train_recall) train_f1score = f1_score(reiss_gold, reiss_needs, average="macro") print("Training f1-score: ", train_f1score) print("-----------------------------------------------------------------------") print("Classification report:") class_report = classification_report(reiss_gold, reiss_needs) print(class_report) cm_reiss = confusion_matrix(reiss_gold, maslow_needs) print(cm_reiss) Loading
Evaluation/train_evaluation.py 0 → 100644 +168 −0 Original line number Diff line number Diff line #!/usr/bin/env python3 # -*- coding: utf-8 -*- """ Created on Fri Mar 26 16:11:25 2021 @author: erkinmenderes """ import pandas as pd #GS_annotations file should be in the same directory, else insert filepath output = pd.read_csv("knowlege_path_subset_train.csv", sep=';', error_bad_lines=False) index_list = output.index.tolist() #output["Essay"] += 1 #print(index_list) essay_output_list = output["Essay"].tolist() essay_output_list = essay_output_list[:94] path_list = output["Path"].tolist() gold_df = pd.read_csv("input_subset_train.csv", sep=';', error_bad_lines=False) gold_essay_list = gold_df["Essay"].tolist() reiss_gold = gold_df["Reiss"].tolist() def replacer(list_string): """ Function to replace all the unnecessary characters in the paths in order to further process the data and return clean strings Parameters ---------- list_string : list a list containing strings of words (here: conceptnet paths as strings) Returns ------- str strings cleaned from unwanted and unnecessary characters and tokens """ text = list_string.replace("[", "").replace("]", "").replace('\'', "").replace("\"", "").replace(",", "") return text.split() # create maslow and reiss human needs maslow_human_needs = ["physiological needs", "stability", "love/belonging", "esteem", "spiritual growth"] reiss_motives = ["food", "rest", "health", "save_money", "order", "safety", "romance", "belonging", "family", "contact", "competition", "honor", "approval", "status", "power", "curiosity", "serenity", "idealism", "independent"] # create a cleaned list of paths cleaned_paths = [replacer(path) for path in path_list] def assign_reiss(path_list): """ Assigns Reiss motive to a a graph consisting of a list of its paths Parameters ---------- path_list : list The entire list of subraphs which consist of their paths paths are split into their single units as strings Returns ------- None. """ temp_list = [] human_needs = [] for path in path_list: for word in path: if word in reiss_motives: temp_list.append(word) human_needs.append(temp_list[0]) temp_list = [] return human_needs #assign reiss human needs for every graph via its top ranked path reiss_needs = assign_reiss(cleaned_paths) #maslow needs list to assign maslow need accordingly physiological_needs = ['food', 'rest'] safety = ['health', 'save_money', 'order', 'safety'] love_belonging = ['love', 'belonging', 'family', 'contact'] esteem = ['competition','honor', 'approval', 'status', 'power'] spiritual_growth = ['curiosity', 'serenity','idealism', 'independent'] def assign_maslow(reiss_list): """ Function that assigns corresponding maslow human need given its reiss human need Parameters ---------- reiss_list : list list containing assigned reiss human need for every essay (via the top ranked graphpath) Returns ------- maslow_needs : list list containing corresponding maslow human needs """ maslow_needs = [] for r in reiss_needs: if r in physiological_needs: maslow_needs.append(maslow_human_needs[0]) elif r in safety: maslow_needs.append(maslow_human_needs[1]) elif r in love_belonging: maslow_needs.append(maslow_human_needs[2]) elif r in esteem: maslow_needs.append(maslow_human_needs[3]) else: maslow_needs.append(maslow_human_needs[4]) return maslow_needs maslow_needs = assign_maslow(reiss_needs) # create joint list of reiss and maslow hn_list_full = list(zip(maslow_needs, reiss_needs)) #post-processing for evaluation in reiss reiss_needs = [w.replace("independent", "independence") for w in reiss_needs] reiss_needs = [w.replace("save_money", "savings") for w in reiss_needs] #post-processing for evaluation in maslow maslow_needs = [w.replace("love / belonging", "love/belonging") for w in maslow_needs] # add columns accordingly output["Maslow_predict"] = maslow_needs output["Reiss_predict"] = reiss_needs from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score from sklearn.metrics import ConfusionMatrixDisplay, classification_report, confusion_matrix train_acc = accuracy_score(reiss_gold, reiss_needs) print("Training accuracy: ", train_acc) train_prec = precision_score(reiss_gold, reiss_needs, average='macro') print("Training precision: ", train_prec) train_recall = recall_score(reiss_gold, reiss_needs, average='macro') print("Training recall: ", train_recall) train_f1score = f1_score(reiss_gold, reiss_needs, average="macro") print("Training f1-score: ", train_f1score) print("-----------------------------------------------------------------------") print("Classification report:") class_report = classification_report(reiss_gold, reiss_needs) print(class_report) cm_reiss = confusion_matrix(reiss_gold, maslow_needs) print(cm_reiss)
