A common problem when creating models to generate business value from data is that the datasets can be so large that it can take days for the model to generate predictions. Ensuring that your dataset is stored as efficiently as possible is crucial for allowing these models to run on a more reasonable timescale without having to reduce the size of the dataset.
You’ve been hired by a major online data science training provider called Training Data Ltd. to clean up one of their largest customer datasets. This dataset will eventually be used to predict whether their students are looking for a new job or not, information that they will then use to direct them to prospective recruiters.
You’ve been given access to customer_train.csv, which is a subset of their entire customer dataset, so you can create a proof-of-concept of a much more efficient storage solution. The dataset contains anonymized student information, and whether they were looking for a new job or not during training:
Column
Description
student_id
A unique ID for each student.
city
A code for the city the student lives in.
city_development_index
A scaled development index for the city.
gender
The student’s gender.
relevant_experience
An indicator of the student’s work relevant experience.
enrolled_university
The type of university course enrolled in (if any).
education_level
The student’s education level.
major_discipline
The educational discipline of the student.
experience
The student’s total work experience (in years).
company_size
The number of employees at the student’s current employer.
company_type
The type of company employing the student.
last_new_job
The number of years between the student’s current and previous jobs.
training_hours
The number of hours of training completed.
job_change
An indicator of whether the student is looking for a new job (1) or not (0).
Task
The Head Data Scientist at Training Data Ltd. has asked you to create a DataFrame called ds_jobs_transformed that stores the data in customer_train.csv much more efficiently. Specifically, they have set the following requirements:
Columns containing categories with only two factors must be stored as Booleans (bool).
Columns containing integers only must be stored as 32-bit integers (int32).
Columns containing floats must be stored as 16-bit floats (float16).
Columns containing nominal categorical data must be stored as the category data type.
Columns containing ordinal categorical data must be stored as ordered categories, and not mapped to numerical values, with an order that reflects the natural order of the column.
The DataFrame should be filtered to only contain students with 10 or more years of experience at companies with at least 1000 employees, as their recruiter base is suited to more experienced professionals at enterprise companies.
Important
If you call .info() or .memory_usage() methods on ds_jobs and ds_jobs_transformed after you’ve preprocessed it, you should notice a substantial decrease in memory usage.
Data Source
Data: The primary data used for this analysis is the customer_train.csv, which is a subset of the entire customer dataset
Tools
Jupyter lab
Steps/Explanations
The necessary library was imported, which is Pandas
The Original dataset was loaded, named ds_jobs and a copy was made, called ds_jobs_transformed
Exploratory Data Analysis was performed which help identify ordinal, nominal, and two-factor categories. This was done by written codes, which iterate over all columns of the DataFrame ds_jobs that have a data type of object (typically representing strings or categorical data) and print the value counts of each column.
A dictionary of columns containing ordered categorical data was created. The code defines the dictionary called ordered_cats, which contains lists of ordered categories for various features. These features represent specific categorical data in a dataset (e.g., levels of education, size of the company, work experience, etc.). This dictionary can later be used to create ordered categorical columns, for example, when transforming or encoding data in a pandas DataFrame.
A mapping dictionary of columns containing two-factor categories to convert to Booleans was created. The code defines a Python dictionary called two_factor_cats. This dictionary is used to map certain categorical values into Boolean (True or False) values.
This for col in ds_jobs_transformed: code iterates through each column in the ds_jobs_transformed DataFrame, performing different transformations based on the column name.
For the columns 'relevant_experience' and 'job_change', if col in ['relevant_experience', 'job_change']: code uses the two_factor_cats dictionary to convert their categorical values into boolean values (True/False).
The .map() function applies the mapping from two_factor_cats (defined previously) to the column. For example:
'No relevant experience' becomes False.
'Has relevant experience' becomes True.
0.0 becomes False.
1.0 becomes True.
For the columns 'student_id' and 'training_hours', elif col in ['student_id', 'training_hours']: code changes their data types to int32 using .astype('int32'). This helps reduce memory usage, especially if the original data type was int64. The smaller int32 uses less memory and is sufficient for storing integers that fit within the 32-bit range.
The column 'city_development_index' is converted to float16 (16-bit floating-point format) by the following code:
elif col in ['student_id', 'training_hours']:
ds_jobs_transformed[col] = ds_jobs_transformed[col].astype('int32')
float16 consumes less memory than the default float64.
It’s useful for reducing the memory footprint of large datasets when the precision provided by float16 is sufficient.
For columns that are in the ordered_cats dictionary created earlier, the code below converts them to ordered categorical data types.
elif col in ordered_cats.keys():
category = pd.CategoricalDtype(ordered_cats[col], ordered=True)
ds_jobs_transformed[col] = ds_jobs_transformed[col].astype(category)
ordered_cats is a dictionary that contains the order of categories for certain columns.
pd.CategoricalDtype creates a categorical data type with a specific order, which is useful when the categories have a meaningful order (e.g., educational levels or experience).
ordered=True ensures that the categories are treated as ordered (e.g., “Primary School” < “High School” < “Graduate”).
The .astype(category) applies this conversion.
For all remaining columns (those not handled in the previous conditions), the code below converts them to the standard categorical data type without an explicit order.
Converting to category reduces memory usage, especially when the column has a limited number of distinct values (e.g., city names, job roles, etc.).
The final DataFrame was filtered to only contain students with 10 or more years of experience at companies with at least 1000 employees, as their recruiter base is suited to more experienced professionals at enterprise companies.
# Import necessary librariesimport pandas as pd# Load the datasetds_jobs = pd.read_csv("customer_train.csv")# View the datasetds_jobs.head()ds_jobs.info()
Include below are the codes used to achieve the task given
# Import necessary librariesimport pandas as pd# Load the dataset and create a copyds_jobs = pd.read_csv("customer_train.csv")ds_jobs_transformed = ds_jobs.copy()# EDA to help identify ordinal, nominal, and two-factor categoriesfor col in ds_jobs.select_dtypes("object").columns:print(ds_jobs_transformed[col].value_counts(), '\n')# Create a dictionary of columns containing ordered categorical dataordered_cats = {'enrolled_university': ['no_enrollment', 'Part time course', 'Full time course'],'education_level': ['Primary School', 'High School', 'Graduate', 'Masters', 'Phd'],'experience': ['<1'] +list(map(str, range(1, 21))) + ['>20'],'company_size': ['<10', '10-49', '50-99', '100-499', '500-999', '1000-4999', '5000-9999', '10000+'],'last_new_job': ['never', '1', '2', '3', '4', '>4']}# Create a mapping dictionary of columns containing two-factor categories to convert to Booleanstwo_factor_cats = {'relevant_experience': {'No relevant experience': False, 'Has relevant experience': True},'job_change': {0.0: False, 1.0: True}}# Loop through DataFrame columns to efficiently change data typesfor col in ds_jobs_transformed:# Convert two-factor categories to boolif col in ['relevant_experience', 'job_change']: ds_jobs_transformed[col] = ds_jobs_transformed[col].map(two_factor_cats[col])# Convert integer columns to int32elif col in ['student_id', 'training_hours']: ds_jobs_transformed[col] = ds_jobs_transformed[col].astype('int32')# Convert float columns to float16elif col =='city_development_index': ds_jobs_transformed[col] = ds_jobs_transformed[col].astype('float16')# Convert columns containing ordered categorical data to ordered categories using dictelif col in ordered_cats.keys(): category = pd.CategoricalDtype(ordered_cats[col], ordered=True) ds_jobs_transformed[col] = ds_jobs_transformed[col].astype(category)# Convert remaining columns to standard categorieselse: ds_jobs_transformed[col] = ds_jobs_transformed[col].astype('category')# Filter students with 10 or more years experience at companies with at least 1000 employeesds_jobs_transformed = ds_jobs_transformed[(ds_jobs_transformed['experience'] >='10') & (ds_jobs_transformed['company_size'] >='1000-4999')]ds_jobs_transformed.info()
There was a marked reduction in the size of the ds_jobs_transformed DataFrame from 2.0+ MB to 69.5 KB, which help to save memory and prepare the data for predictive modelling.
Recommendations
None
Limitations
None
References
Filtering DataFrames in Intermediate Python Course for Associate Data Scientist in Python Carrer Track in DataCamp Inc by Hugo Bowne-Henderson.
For loop in Intermediate Python Course for Associate Data Scientist in Python Carrer Track in DataCamp Inc by Hugo Bowne-Henderson.
Python For Data Analysis 3E (Online) by Wes Mckinney Click here to preview
Working with Categorical Data in Python in Intermediate Python Course for Associate Data Scientist in Python Carrer Track in DataCamp Inc by Kasey Jones.
