US Airline Tweet Sentiment

import pandas as pd

import matplotlib.pyplot as plt

import numpy as np

import geopandas as gpd

from shapely.geometry import Point,Polygon

These are the basic libraries required to perform data visualization task. We can also use seaborn or plotly to make our plots more aesthetic.

To install shapely : pip install shapely

To install geopandas refer to this documentation: link

data=pd.read_csv('tweet.csv')

data.head()

data.shape

data.isnull().any()

data['airline'].value_counts()

These are also the fundamental measures to know the data and to ensure that the dataset has no NULL values. If it has any null values then we can use data preprocessing techniques to remove the NAN's, missing values and outliers.

plt.figure(figsize=(15,10))

plt.subplot(331)

data[data['airline']=='Delta']['airline_sentiment'].value_counts().plot(kind='barh',color='teal')

plt.title('delta airline')

plt.subplot(332)

data[data['airline']=='American']['airline_sentiment'].value_counts().plot(kind='barh',color='darkturquoise')

plt.title('American airline')

plt.subplot(333)

data[data['airline']=='United']['airline_sentiment'].value_counts().plot(kind='barh',color='cadetblue')

plt.title('united airline')

plt.subplot(334)

data[data['airline']=='US Airways']['airline_sentiment'].value_counts().plot(kind='barh',color='powderblue')

plt.title('US Airways')

plt.subplot(335)

data[data['airline']=='Southwest']['airline_sentiment'].value_counts().plot(kind='barh',color='mediumaquamarine')

plt.title('Southwest airline')

plt.subplot(336)

data[data['airline']=='Virgin America']['airline_sentiment'].value_counts().plot(kind='barh',color='steelblue')

plt.title('Virgin America')

We have used subplots to place various bar graphs side by side. Barh is used for the horizontal bar graph. If we want to plot a vertical bar graph then we can use only bar . value_counts() gives us the unique values in that particular attribute and also keep a count on the number of unique values in a given attribute.

geo=[Point(xy) for xy in zip(data[data['airline']=='Delta']['longitude'][:200],data[data['airline']=='Delta']['latitude'][:200])]

geo1=[Point(xy) for xy in zip(data[data['airline']=='Southwest']['longitude'][:200],data[data['airline']=='Southwest']['latitude'][:200])]

geo2=[Point(xy) for xy in zip(data[data['airline']=='American']['longitude'][:200],data[data['airline']=='American']['latitude'][:200])]

geo3=[Point(xy) for xy in zip(data[data['airline']=='United']['longitude'][:200],data[data['airline']=='United']['latitude'][:200])]

geo4=[Point(xy) for xy in zip(data[data['airline']=='US Airways']['longitude'][:200],data[data['airline']=='US Airways']['latitude'][:200])]

geo5=[Point(xy) for xy in zip(data[data['airline']=='Virgin America']['longitude'][:200],data[data['airline']=='Virgin America']['latitude'][:200])]

We first will convert latitude and longitude into Point data type. We have grouped similar airlines and found out the location of tweets based on different airlines.

g=gpd.GeoDataFrame(geometry=geo)

g1=gpd.GeoDataFrame(geometry=geo1)

g2=gpd.GeoDataFrame(geometry=geo2)

g3=gpd.GeoDataFrame(geometry=geo3)

g4=gpd.GeoDataFrame(geometry=geo4)

g5=gpd.GeoDataFrame(geometry=geo5)

We now will convert the points into geopandas data frame in order to plot those plots on the US map.

US = gpd.read_file('States_shapefile-shp/States_shapefile.shp')

You can download the US shape file from here. A shapefile is a simple, nontopological format for storing the geometric location and attribute information of geographic features.

fig,ax=plt.subplots(figsize=(15,10))

US.boundary.plot(ax=ax,color='slategrey')

g.plot(ax=ax,color='lightblue',label='delta')

g1.plot(ax=ax,color='plum',label='southwest')

g2.plot(ax=ax,color='mediumslateblue',label='American')

g3.plot(ax=ax,color='darkgoldenrod',label='United')

g4.plot(ax=ax,color='lightcoral',label='US Airways')

g5.plot(ax=ax,color='forestgreen',label='Virgin America')

plt.legend()

This is how we will plot the points. We first will create and figure and axes. On that axes, we will overlap out geometric points.