Patterns over time

When data includes dates or time values, we'll want to examine whether there might be patterns over time

Importing DateTime data

df = pd.read_csv('datasets.csv', parse_dates=['birth_date'])

Converting to DateTime data

df['birth_date'] = pd.to_datetime(df['birth_date'])

Creating DateTime data

# Column must be month, date, year
df['birth_date'] = pd.to_datetime(['month', 'date', 'year'])

Extracting DateTime data

df['month'] = df['birth_date'].dt.month
df['date'] = df['birth_date'].dt.day
df['year'] = df['birth_date'].dt.year

Visualizing patterns over time

sns.lineplot(x='marriage_month', y='marriage_duration', data=df)
plt.show()

Importing DateTime data

# Import divorce.csv, parsing the appropriate columns as dates in the import
divorce = pd.read_csv('divorce.csv', parse_dates=['divorce_date', 'marriage_date', 'dob_man', 'dob_woman'])
print(divorce.dtypes)

Updating data type to DateTime

# Convert the marriage_date column to DateTime values
divorce["marriage_date"] = pd.to_datetime(divorce['marriage_date'])

Visualizing relationships over time

# Define the marriage_year column
divorce["marriage_year"] = divorce["marriage_date"].dt.year

# Create a line plot showing the average number of kids by year
sns.lineplot(x='marriage_year', y='num_kids', data=divorce)
plt.show()

Correlation

• Describes the direction and strength of relationships between two variables

• Help us use variables to predict future outcomes

• .corr() calculates the Pearson correlation coefficient, measuring linear relationships

Visualizing correlation

sns.heatmap(df.corr(), annot=True)
plt.show()

sns.pairplot(data=divorce)
plt.show()

# Limit variables
sns.pairplot(data=divorce, vars=['income_man', 'income_woman', 'marriage_duration']
plt.show()

Visualizing variable relationships

# Create the scatterplot
sns.scatterplot(x='marriage_duration', y='num_kids', data=divorce)
plt.show()

Visualizing multiple variable relationships

# Create a pairplot for income_woman and marriage_duration
sns.pairplot(data=divorce,vars=['income_woman', 'marriage_duration'])
plt.show()

Factor relationships and distributions

Categorial data in scatter plots

# Create the scatter plot
sns.scatterplot(x='woman_age_marriage', y='income_woman', data=divorce, hue='education_woman')
plt.show()

Exploring with KDE plots

# Update the KDE plot to show a cumulative distribution function
sns.kdeplot(data=divorce, x="marriage_duration", hue="num_kids", cut=0, cumulative=True)
plt.show()