Time Series Analysis and Visualization of FB Comments

The representation of data that is collected over time helps to identify trends, patterns, and anomalies within the data. By plotting data points along a timeline, you can visually analyze the movement of values over time.

Exhibit 25.22 provides the Python code for the visualization of time series data. The example uses data stored in the file fb_comments_metrics.csv, which contains information on the likes and shares of comments on Facebook posts throughout a day.

The Seaborn library is used to create the bar plots in this example. Built on top of Matplotlib, Seaborn provides a high-level interface for creating attractive and informative statistical graphics. For more information on Seaborn and to explore additional examples of its use in statistical analysis and visualization, refer to the section Seaborn Visualization in Python.  

The code in Exhibit 25.22 is a continuation of our analysis of Facebook textual data, demonstrating how representing data collected over time can highlight trends, patterns, and anomalies.

The code analyzes data about likes and shares received by Facebook posts throughout a day. Here’s a breakdown of the steps in the analysis:

1. Import libraries: Import the tools needed to work with the data (like math functions, data frames, and plotting tools).
2. Read data: Read information about likes and shares from a file called fb_comments_metrics.csv.
3. Prepare the data for analysis: Set the size and font size for the charts to be clear and easy to read. Extract the time information (hour, day, month, year) from the existing “created_time” column to analyze how metrics change over time.
4. Likes by Hour: Create a bar chart where the x-axis shows the hours of the day (1-24) and the y-axis shows the average number of likes (blue bars) received for each hour.
5. Shares by Hour: Repeat step 4 for the number of shares (green bars).
6. Shares by Day: Set the background of the chart to white and create a bar chart where the x-axis shows the days of the month (1-31) and the y-axis shows the average number of shares (black bars)received for each day.

In short, this code helps you see how the number of likes and shares for Facebook posts change throughout the day and across different days.

import matplotlib.pyplot as plt
import seaborn as sns
import numpy as np
import pandas as pd

df = pd.read_csv(r"data/fb_comments_metrics.csv", encoding ="latin-1") 
'''
`r` prefix preceding a string literal denotes a raw string literal.
Specifically, backslashes (\) are treated as literal characters, and are not 
used for escaping special characters like they are in regular string literals.
'''
# set figure size and font size
sns.set(rc={'figure.figsize':(40,20)})
sns.set(font_scale=3)

# Separating time by Hour, Day, Month and Year for further analysis using datetime package
import datetime as dt
'''
A lambda function is a small anonymous function that can take any number of arguments, but can only have one expression.
e.g.:
lambda x: x.hour
paramter: x
expression: x.hour ... converts time in datetime format to hour and returns hour
'''
df['time'] = pd.to_datetime(df['created_time'])
df['hour'] = df['time'].apply(lambda x: x.hour)
df['month'] = df['time'].apply(lambda x: x.month)
df['day'] = df['time'].apply(lambda x: x.day)
df['year'] = df['time'].apply(lambda x: x.year)
df.head()

# set x labels
x_labels = df.hour

#create bar plot
sns.barplot(x=x_labels, y=df.likes, color="blue")

# display the plot
plt.show()

# only show x-axis labels for Jan 1 of every other year
# tick_positions = np.arange(10, len(x_labels), step=24)

#create bar plot
sns.barplot(x=x_labels, y=df.shares, color="green")

# display the plot
plt.show()

# Set the background color to white
sns.set_style(rc = {'axes.facecolor': 'white'})

#create bar plot
sns.barplot(x=df.day, y=df.shares, color="black")

# display the plot
plt.show()

Exhibit 25.22   Seaborn visualization of time series: This analysis of the likes and share of Facebook posts demonstrates how to generate bar plots of the metrics using Seaborn. Jupyter notebook.