Market Analysis (Python)

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
sns.set(style='white', palette='Pastel2')
import os

def read_file(filepath, plot = True):
    """
    Read a CSV file from a given filepath, convert it into a DataFrame, and return a
    processed DataFrame with three columns: 'week', 'region', and 'interest'. 
    Generate a line plot using Seaborn to visualize the data. 
    This corresponds to the first graphic (time series) returned by trends.google.com. 
    """
    if not hasattr(read_file, "count"): # if the function has not been called before, set the count to 1
        read_file.count = 1
    else: # if the function has been called before, increment the count by 1
        read_file.count += 1

    file = pd.read_csv(filepath, header=1)
    df = file.set_index('Week').stack().reset_index()
    df.columns = ['week','region','interest']
    df['week'] = pd.to_datetime(df['week'])
    plt.figure(figsize=(8,3))
    df = df[df['interest']!="<1"]
    df['interest'] = df['interest'].astype(float)

    if plot: # modify the code so it gives the plot a title "Figure [n]: Google Trends search ratio over time". n is how many times the function has been called.
        plt.title(f"Figure {read_file.count}: Google Trends search ratio over time")
        sns.lineplot(data = df, x= 'week', y= 'interest', hue='region')
    return df

def read_geo(filepath, multi=False):
    """
    Read a CSV file from a given filepath, convert it into a pandas DataFrame,
    and return a processed DataFrame with two columns: 'country' and 'interest'. 
    Generate a bar plot using Seaborn to visualize the data. 
    This corresponds to the second graphic returned by trends.google.com. 
    Use multi=False if only one keyword is being analyzed, and 
    multi=True if more than one keyword is being analyzed.
    """
    file = pd.read_csv(filepath, header=1)

    if not multi:
        file.columns = ['country', 'interest']
        plt.figure(figsize=(8,4))
        sns.barplot(data = file.dropna().iloc[:25,:], y = 'country', x='interest')

    if multi:
        plt.figure(figsize=(3,8))
        file = file.set_index('Country').stack().reset_index()
        file.columns = ['country','category','interest']
        file['interest'] = pd.to_numeric(file['interest'].apply(lambda x: x[:-1]))
        sns.barplot(data=file.dropna(), y = 'country', x='interest', hue='category')

    file = file.sort_values(ascending=False,by='interest')
    return file

df=read_file("data/workout.csv")
dataset_name = "workout.csv"

import pandas as pd
import scipy.stats as stats

def generate_report(df, dataset_name):
    num_duplicate_rows = df.duplicated().sum()
    report = f"The dataset in {dataset_name} has {df.shape[0]} rows "
    if num_duplicate_rows == 0:
        report += "from which 0 are duplicates."
    else:
        report += f"from which {num_duplicate_rows} are duplicates.\n"
    
    report += "\n\nThe Data is observed via {} columns:\n".format(df.shape[1])
    
    for column in df.columns:
        report += get_column_info(df, column)
    
    report += additional_data_validation_tasks(df)
    
    return report

def get_column_info(df, column_name):
    num_unique_values = df[column_name].nunique()
    num_missing_values = df[column_name].isnull().sum()
    
    missing_status = f"and {num_missing_values} missing values."
    
    info = f"- {column_name}: Contained {num_unique_values} unique values {missing_status}\n"
    
    return info

def additional_data_validation_tasks(df):
    # Outlier Detection Task
    outlier_status = detect_outliers(df)

    # Skewness and Kurtosis Task
    skewness_kurtosis_status = calculate_skewness_kurtosis(df)
    
    return outlier_status + skewness_kurtosis_status

def detect_outliers(df):
    # You can implement your outlier detection logic here.
    # For example, identify and report columns with potential outliers.
    
    # Placeholder logic for illustration purposes
    outlier_columns = [col for col in df.columns if df[col].dtype == 'float64']
    outlier_status = f"\nThe following columns contained outliers: {', '.join(outlier_columns)}\n"
    
    return outlier_status

def calculate_skewness_kurtosis(df):
    skewness_kurtosis_status = ""
    
    for col in df.select_dtypes(include=['float64']).columns:
        skewness = stats.skew(df[col].dropna())
        kurtosis = stats.kurtosis(df[col].dropna())
        skewness_kurtosis_status += f"{col}: Skewness = {skewness:.2f}, Kurtosis = {kurtosis:.2f}\n"
    
    return skewness_kurtosis_status

# Call the generate_report function to generate the report
report_text = generate_report(df, dataset_name)

# Print the report
print(report_text)

keywords = read_file('data/three_keywords.csv')

current = 'workout'
peak_covid = 'home workout'