Understanding how EDA is done in Python
Various steps involved in the Exploratory Data Analysis
Performing EDA on a given dataset
Exploratory data analysis popularly known as EDA is a process of performing some initial investigations on the dataset to discover the structure and the content of the given dataset. It is often known as Data Profiling. It is an unavoidable step in the entire journey of data analysis right from the business understanding part to the deployment of the models created.
EDA is where we get the basic understanding of the data in hand which then helps us in the further process of Data Cleaning & Data Preparation.
We will be covering a wide range of topics under EDA starting from the basic data exploration (structure based) to the normalization and the standardization of the data. In this article, we will be using the Python programming language to perform the EDA steps.
Let’s see what all we are going to cover!
Introducing the Dataset
Importing the Python Libraries
Loading the Dataset in Python
Structured Based Data Exploration
Handling Duplicates
Handling Outliers
Handling Missing Values
Univariate Analysis
Bivariate Analysis
For this article, we will be using the Black Friday dataset which can be downloaded from here.
Let’s import all the python libraries we will be needing for our analysis namely NumPy, Pandas, Matplotlib and Seaborn.
Now let’s load our dataset into Python. We will be reading the data from a CSV (comma-separated values) file into a Pandas DataFrame naming it as df here.
Python Code:
Let’s begin with the basic exploration of the data we have!
It is the very first step in EDA which can also be referred to as Understanding the MetaData! That’s correct, ‘Data about the Data’.
It is here that we get the description of the data we have in our data frame.
Let’s try now.
Display the FIRST 5 Observations
Display the Number of Variables & Number of Observations
/>df.shape() gives us a tuple having 2 values.
df.dtypes
This gives us the type of variables in our dataset.
Count the Number of Non-Missing Values for each Variable
df.count()
This gives the number of non-missing values for each variable and is extremely useful while handling missing values in a data frame.
Now to know about the characteristics of the data set we will use the df.describe() method which by default gives the summary of all the numerical variables present in our data frame.
df.describe()
Using the df.describe() method we get the following characteristics of the numerical variables namely to count (number of non-missing values), mean, standard deviation, and the 5 point summary which includes minimum, first quartile, second quartile, third quartile, and maximum.
What about the categorical variables?
df.describe(include = 'all')
By providing the include argument and assigning it the value ‘all’ we get the summary of the categorical variables too. For the categorical variables, we get the characteristics: count (number of non-missing values) , unique (number of unique values), top (the most frequent value), and the frequency of the most frequent value.
df.info()
By just this one command of df.info() we get the complete information of the data in hand.
With this, we are done with the Structure-Based Exploratory Data Analysis and now it’s time to get into the Content Based Exploratory Data Analysis.
This involves 2 steps: Detecting duplicates and Removing duplicates.
To check for the duplicates in our data
df.duplicated()
Hereby duplicates mean the exact same observations repeating themselves. As we can see that there are no duplicate observations in our data and hence each observation is unique.
However, to remove the duplicates(if any) we can use the code :
df.drop_duplicates()
Further, we can see that there are duplicate values in some of the variables like User_ID. How can we remove those?
df.drop_duplicates(subset='User_ID')
This by default keeps just the first occurrence of the duplicated value in the User_ID variable and drops the rest of them. Hold On! Here we do not want to remove the duplicate values from the User_ID variable permanently so just to see the output and not make any permanent change in our data frame we can write the command as:
df.drop_duplicates(subset='User_ID' , inplace=False)
As we can see, the values in the User_ID variable are all unique now.
So this is how detection and removal of duplicated observations/values are done in a data frame.
What are Outliers? Outliers are the extreme values on the low and the high side of the data. Handling Outliers involves 2 steps: Detecting outliers and Treatment of outliers.
Detecting Outliers
For this we consider any variable from our data frame and determine the upper cut off and the lower cutoff with the help of any of the 3 methods namely :
Let’s consider the Purchase variable. Now we will be determining if there are any outliers in our data set using the IQR(Interquartile range) Method. What is this method about? You will get to know about it as we go along the process so let’s start. Finding the minimum(p0), maximum(p100), first quartile(q1), second quartile(q2), the third quartile(q3), and the iqr(interquartile range) of the values in the Purchase variable.
p0=df.Purchase.min() p100=df.Purchase.max() q1=df.Purchase.quantile(0.25) q2=df.Purchase.quantile(0.5) q3=df.Purchase.quantile(0.75) iqr=q3-q1
Now since we have all the values we need to find the lower cutoff(lc) and the upper cutoff(uc) of the values.
lc = q1 - 1.5*iqr uc = q3 + 1.5*iqr
lc
uc
We have the uppercut off and the lower cutoff, what now? We will be using the convention :
If lc < p0 → There are NO Outliers on the lower side
If uc > p100 → There are NO Outliers on the higher side
print( "p0 = " , p0 ,", p100 = " , p100 ,", lc = " , lc ,", uc = " , uc)
Clearly lc < p0 so there are no outliers on the lower side. But uc < p100 so there are outliers on the higher side. We can get a pictorial representation of the outlier by drawing the box plot.
df.Purchase.plot(kind='box')
Now since we have detected the outliers it is time to treat those.
Do not worry about the data loss as here we are not going to remove any value from the variable but rather clip them. In this process, we replace the values falling outside the range with the lower or the upper cutoff accordingly. By this, the outliers are removed from the data and we get all the data within the range.
Clipping all values greater than the upper cutoff to the upper cutoff :
df.Purchase.clip(upper=uc)
To finally treat the outliers and make the changes permanent :
df.Purchase.clip(upper=uc,inplace=True) df.Purchase.plot(kind='box')
What are Missing Values? Missing Values are the unknown values in the data. This involves 2 steps: Detecting the missing values and Treatment of the Missing Values
Detecting the Missing Values
df.isna()
df.isna() returns True for the missing values and False for the non-missing values.
Here we are going to find out the percentage of missing values in each variable.
df.isna().sum()/df.shape[0]
And we get from the output that we do have missing values in our data frame in 2 variables: Product_Category_2 and Product_Category_3, so detection is done.
To treat the missing values we can opt for a method from the following :
For variable Product_Category_2, 31.56% of the values are missing. We should not drop such a large number of observations nor should we drop the variable itself hence we will go for imputation. Data Imputation is done on the Series. Here we replace the missing values with some value which could be static, mean, median, mode, or an output of a predictive model.
Since it is a categorical variable, let’s impute the values by mode.
df.Product_Category_2.mode()[0] df.Product_Category_2.fillna(df.Product_Category_2.mode()[0],inplace=True)
Done!
df.isna().sum()
For variable Product_Category_3, 69.67% of the values are missing which is a lot hence we will go for dropping this variable.
df.dropna(axis=1,inplace=True)
df.dtypes
Analysis using Charts
In this type of analysis, we use a single variable and plot charts on it. Here the charts are created to see the distribution and the composition of the data depending on the type of variable namely categorical or numerical.
For Continuous Variables: To see the distribution of data we create Box plots and Histograms.
Distribution of Purchase
Histogram
df.Purchase.hist() plt.show()
We created this histogram using the hist() method of the Series but there is another method too known as plot() by which we can create many more charts.
df.Purchase.plot(kind='hist' , grid = True) plt.show()
We have another way to create this chart by directly using matplotlib!
plt.hist(df.Purchase) plt.grid(True) plt.show()
Box Plot
df.Purchase.plot(kind='box') plt.show()
plt.boxplot(df.Purchase) plt.show()
For Categorical Variables :
Composition of Gender
df.groupby('Gender').City_Category.count().plot(kind='pie') plt.show()
Distribution of Marital_Status
sns.countplot(df.Marital_Status) plt.show()
Composition of City_Category
df.groupby('City_Category').City_Category.count().plot(kind='pie') plt.show()
Distribution of Age
sns.countplot(df.Age) plt.show()
Composition of Stay_In_Current_City_Years
df.groupby('Stay_In_Current_City_Years').City_Category.count().plot(kind='pie') plt.show()
Distribution of Occupation
sns.countplot(df.Occupation) plt.show()
Distribution of Product_Category_1
df.groupby('Product_Category_1').City_Category.count().plot(kind='barh') plt.show()
In this type of analysis, we take two variables at a time and create charts on them. Since we have 2 types of variables Categorical and Numerical so there can be 3 cases in bivariate analysis :
Numerical & Numerical: To see the relationship between the 2 variables we create Scatter Plots and a Correlation Matrix with a Heatmap on the top.
Scatter Plot
Since there is only 1 numerical variable in our dataset so we cannot create the Scatter plot here. But how can we do so? Let’s take a hypothetical example such that we consider all the numeric variables(having dtype as int or float) here as numerical variables.
Considering 2 categorical variables Product_Category_1 and Product_Category_2
df.plot(x='Product_Category_1',y='Product_Category_2',kind = 'scatter') plt.show()
plt.scatter(x=df.Product_Category_1 , y=df.Product_Category_2) plt.show()
Finding a correlation between all the numeric variables.
df.select_dtypes(['float64' , 'int64']).corr()
Creating a heatmap using Seaborn on the top of the correlation matrix obtained above to visualize the correlation between the different numerical columns of the data. This is done when we have a large number of variables.
sns.heatmap(df.select_dtypes(['float64' , 'int64']).corr(),annot=True) plt.show()
Numerical & Categorical
Comparison between Purchase and Occupation: Bar Chart
df.groupby('Occupation').Purchase.sum().plot(kind='bar') plt.show()
summary=df.groupby('Occupation').Purchase.sum() plt.bar(x=summary.index , height=summary.values) plt.show()
sns.barplot(x=summary.index , y=summary.values) plt.show()
Comparison between Purchase and Age: Line Chart
df.groupby('Age').Purchase.sum().plot(kind='line') plt.show()
Composition of Purchase by Gender: Pie Chart
df.groupby('Gender').Purchase.sum().plot(kind='pie') plt.show()
Comparison between Purchase and City_Category: Area Chart
df.groupby('City_Category').Purchase.sum().plot(kind='area') plt.show()
Comparison between Purchase and Stay_In_Current_City_Years: Horizontal Bar Chart
df.groupby('Stay_In_Current_City_Years').Purchase.sum().plot(kind='barh') plt.show()
Comparison between Purchase and Marital_Status
sns.boxplot(x='Marital_Status',y='Purchase',data=df) plt.show()
Categorical & Categorical: To see the relationship between the 2 variables we create a crosstab and a heatmap on top.
Relationship between Age and Gender: Creating a crosstab showing the date for Age and Gender
pd.crosstab(df.Age,df.Gender)
Heatmap: Creating a Heat Map on the top of the crosstab.
sns.heatmap(pd.crosstab(df.Age,df.Gender)) plt.show()
Relationship between City_Category and Stay_In_Current_City_Years
sns.heatmap(pd.crosstab(df.City_Category,df.Stay_In_Current_City_Years)) plt.show()
Finally, we have come to the end of this article. In this article, we took a sample data set and performed exploratory data analysis on it using the Python programming language using the Pandas DataFrame. However, this was just a basic idea on how EDA is done you can definitely explore it to as much extent as you want and try performing the steps on bigger datasets as well.
Read more articles on our blog.
You can connect with me on LinkedIn.