Kolawole Jegede

Sentiment Analysis of the British Airways Customers' Feedback Data Overview In [3]: # import the libraries and data import pandas as pd import numpy as np import nltk import matplotlib.pyplot as plt %matplotlib inline import seaborn as sns import re # set plot style sns.set_theme() In [4]: # import the dataset df = pd.read_csv("reviewdata.csv") df.info() RangeIndex: 2500 entries, 0 to 2499 Data columns (total 10 columns): # Column Non-Null Count --- ------------------0 verify 2092 non-null 1 review 2475 non-null 2 date 2500 non-null 3 country 2500 non-null 4 seat_type 2500 non-null 5 recommended 2500 non-null 6 stars 2500 non-null 7 departure 2500 non-null 8 arrival 2500 non-null 9 type_of_traveller 2500 non-null dtypes: int64(1), object(9) memory usage: 195.4+ KB In [5]: # drop empty rows in the review column df = df.dropna(subset = ["review"]) df.info() Dtype ----object object object object object object int64 object object object Index: 2475 entries, 0 to 2499 Data columns (total 10 columns): # Column Non-Null Count --- ------------------0 verify 2067 non-null 1 review 2475 non-null 2 date 2475 non-null 3 country 2475 non-null 4 seat_type 2475 non-null 5 recommended 2475 non-null 6 stars 2475 non-null 7 departure 2475 non-null 8 arrival 2475 non-null 9 type_of_traveller 2475 non-null dtypes: int64(1), object(9) memory usage: 212.7+ KB Dtype ----object object object object object object int64 object object object In [6]: df.head() Out[6]: 0 1 2 3 4 verify review date country seat_type recommended stars departure Trip Verified I had the most fantastic BA Flight today. The ... 1st August 2023 Hong Kong Business CLas Vegass yes 5 Heathrow La Trip Verified Couldn't book in online. Arrived at check in t... 31st July 2023 United Kingdom Economy CLas Vegass no 3 Rome He Trip Verified London Heathrow to Mumbai in a Boeing 787-8 in... 31st July 2023 Iceland Business CLas Vegass yes 3 Gatwick Trip Verified Keflavík, Iceland to London Heathrow on an A32... 31st July 2023 Iceland Business CLas Vegass yes 5 London Trip Verified Terrible Experience with British Airways. I bo... 29th July 2023 Canada Economy CLas Vegass no 5 Denver He In [7]: # correct the class types in the seat_type columns df["seat_type"] = df["seat_type"].replace({"Business CLas Vegass": "Business Cla df["seat_type"].unique() # make the corrections in the review column also df["review"] = df["review"].str.replace(r"(?i)clas vegass", "Class", regex=True) Basic Data Exploration In [9]: # Explore review by seat type df["seat_type"].value_counts() Out[9]: seat_type Economy Class 1351 Business Class 745 Premium Economy 245 First Class 134 Name: count, dtype: int64 In [10]: # a plot showing the seat types plt.figure(figsize=(10, 6)) sns.countplot(y=df["seat_type"], order=df["seat_type"].value_counts().index, pal plt.xlabel("Count") plt.ylabel("Seat Class") plt.title("Distribution of Seat Classes") plt.show() In [11]: #Explore data by traveller type df["type_of_traveller"].value_counts() Out[11]: type_of_traveller Couple Leisure 991 Family Leisure 499 Business 493 Solo Leisure 492 Name: count, dtype: int64 In [12]: # a plot showing the different types of travellers plt.figure(figsize=(10, 6)) sns.countplot(y=df["type_of_traveller"], order=df["type_of_traveller"].value_cou plt.xlabel("Count") plt.ylabel("Type of Traveller") plt.title("Distribution of Traveller Types") plt.show() In [13]: # the most frequent destination df["arrival"].value_counts() Out[13]: arrival Heathrow LHR Las Vegas Luanda Tampa Sydney Venice CPT Name: count, - dtype: int64 In [14]: # a plot showing the different destinations plt.figure(figsize=(10, 6)) sns.countplot(y=df["arrival"], order=df["arrival"].value_counts().index, palette plt.xlabel("Count") plt.ylabel("Different Destinations") plt.title("Distribution of The Most Frequent Destinations") plt.show() Data Preparation for Key Topic Extraction and Sentiment Analysis In [16]: ## Preprocess text for analysis from nltk.corpus import stopwords nltk.download("stopwords") stop_words = set(stopwords.words("english")) [nltk_data] Downloading package stopwords to [nltk_data] C:\Users\User\AppData\Roaming\nltk_data... [nltk_data] Package stopwords is already up-to-date! In [17]: # Create a preprocess function def preprocess_text(text): text = re.sub(r'\W', ' ', str(text)) # to remove special characters text = text.lower() # to convert to lowercase text = " ".join(word for word in text.split() if word not in stop_words) return text In [18]: # replace review column with a preprocessed column df["cleaned_review"] = df["review"].apply(preprocess_text) Key Topic Extraction In [20]: # import vectorizer and LDA from sklearn.feature_extraction.text import CountVectorizer from sklearn.decomposition import LatentDirichletAllocation In [21]: # converting text to document-term matrix vectorizer = CountVectorizer(max_df=0.9, min_df=2, stop_words='english') doc_term_matrix = vectorizer.fit_transform(df["cleaned_review"]) # In [22]: # LDA for topic extraction lda = LatentDirichletAllocation(n_components=5, random_state=42) lda.fit(doc_term_matrix) Out[22]: ▾ LatentDirichletAllocation LatentDirichletAllocation(n_components=5, random_state=42) In [23]: # Display top words for each topic words = vectorizer.get_feature_names_out() for i, topic in enumerate(lda.components_): print(f"Topic {i+1}: ", [words[j] for j in topic.argsort()[-10:]]) Topic 1: ['london', 'cabin', 'class', 'good', 'seat', 'crew', 'service', 'food', 'ba', 'flight'] Topic 2: ['booked', 'hours', 'told', 'service', 'customer', 'airways', 'britis h', 'london', 'ba', 'flight'] Topic 3: ['london', 'customer', 'pay', 'airways', 'british', 'heathrow', 'servic e', 'seat', 'ba', 'airline'] Topic 4: ['baggage', 'london', 'seats', 'ba', 'staff', 'bag', 'flight', 'boardin g', 'luggage', 'check'] Topic 5: ['got', 'said', 'plane', 'flight', 'seats', 'luggage', 'asked', 'londo n', 'ba', 'staff'] In [24]: ## Display Topics in a Word Cloud In [25]: from wordcloud import WordCloud In [26]: # Loop through each topic for i, topic in enumerate(lda.components_): # Create a dictionary of words and their importance word_freq = {words[j]: topic[j] for j in topic.argsort()[-20:]} # Top 20 wo # Generate word cloud wordcloud = WordCloud(width=800, height=400, background_color='white').gener # Plot the word cloud plt.figure(figsize=(8, 4)) plt.imshow(wordcloud, interpolation="bilinear") plt.axis("off") plt.title(f"Topic {i+1}") plt.show() Topics by Arrival Region In [28]: # Convert reviews into a document-term matrix using your vectorizer review_matrix = vectorizer.transform(df['cleaned_review']) # Predict topics for each review topic_probabilities = lda.transform(review_matrix) df['dominant_topic'] = topic_probabilities.argmax(axis=1) # Count topics per destination topic_counts = df.groupby(['arrival', 'dominant_topic']).size().reset_index(name # Pivot table for heatmap pivot_table = topic_counts.pivot(index='arrival', columns='dominant_topic', valu # Plot heatmap plt.figure(figsize=(12, 6)) sns.heatmap(pivot_table, cmap="Blues", annot=True, fmt=".0f") plt.title("Topic Distribution by Arrival Destination") plt.xlabel("Topic") plt.ylabel("Arrival Destination") plt.show() Topics by Seat Class In [30]: # Count topics per seat class topic_counts = df.groupby(['seat_type', 'dominant_topic']).size().reset_index(na # Pivot table for heatmap pivot_table = topic_counts.pivot(index='seat_type', columns='dominant_topic', va # Plot heatmap plt.figure(figsize=(12, 6)) sns.heatmap(pivot_table, cmap="Blues", annot=True, fmt=".0f") plt.title("Topic Distribution by Seat Class") plt.xlabel("Topic") plt.ylabel("Seat Class") plt.show() In [ ]: