Scribd
Upload
5 views2 pages

Counterfeit Medicine Sales Prediction Model

The document outlines a project focused on predicting sales figures for counterfeit medicines, which pose significant health risks and are prevalent in developing countries. It includes a formal problem statement, data files for training and testing, and evaluation criteria based on mean absolute error (MAE). The project is divided into two parts: a quiz based on data exploration and the creation of machine learning models to minimize MAE for successful submission.

Uploaded by

Hem Kuniyal
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
5 views2 pages

Counterfeit Medicine Sales Prediction Model

The document outlines a project focused on predicting sales figures for counterfeit medicines, which pose significant health risks and are prevalent in developing countries. It includes a formal problem statement, data files for training and testing, and evaluation criteria based on mean absolute error (MAE). The project is divided into two parts: a quiz based on data exploration and the creation of machine learning models to minimize MAE for successful submission.

Uploaded by

Hem Kuniyal
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

16/01/2019 Project3

Counterfeit Medicines Sales Prediction

Counterfeit medicines are fake medicines which are either contaminated or contain wrong or no active
ingredient. They could have the right active ingredient but at the wrong dose. Counterfeit drugs are illegal and
are harmful to health. 10% of the world's medicine is counterfeit and the problem is even worse in developing
countries. Up to 30% of medicines in developing countries are counterfeit.

Millions of pills, bottles and sachets of counterfeit and illegal medicines are being traded across the world. The
World Health Organization (WHO) is working with International Criminal Police Organization (Interpol) to dislodge
the criminal networks raking in billions of dollars from this cynical trade.

Despite all these efforts, counterfeit medicine selling rackets don’t seem to stop popping here and there. It has
become a challenge to deploy resources to counter these; without spreading them too thin and eventually
rendering them ineffective. Government has decided that they should focus on illegal operations of high net
worth first instead of trying to control all of them. In order to do that they have collected data which will help them
to predict sales figures given an illegal operation's characteristics.

Data Files

Train Dataset = counterfeit_train.csv

Test Dataset = counterfeit_test.csv

Formal Problem Statement

Variable names are self explanatory.

Your task here is to build a predictive model for predicting sales figures given other information related to
counterfeit medicine selling operations. You need to build your model on the train dataset. Test dataset does not
have a response column; you need to predict those values and submit it in a csv format.

Evaluation Criterion

Part 1:

[Link] 1/2
16/01/2019 Project3

You will first attempt Part 1 of this project which is a quiz. You can access it through LMS. This quiz needs to be
answered based on exploration of the dataset given and some generic questions about algorithms discussed in
the course. Consider only the training dataset for data cleaning and exploration to answer the quiz questions.
There will be 10 questions of which you need to get at least 7 correct in order to pass the project.

Part 2:

Here you work on creating the machine learning models and choosing the one which gives the best
performance. You can refer to the Project Process Guides provided in LMS to understand how to approach and
work on a project.

For this project, score will be calculated as:

Score = 1-(MAE/1660)

where MAE is mean absolute error on test file. You need to score more than 0.5 in order to pass the project
submission. Don't read too much into score formulation, it is just to scale MAE. You just need to focus on
minimizing MAE.

Submission:

Submission CSV should resemble the file:

Sample Submission = 'sample_submission.csv'

Column names, value types should be exactly the same. Also number of rows in the submission csv should be
exactly the same as test data. If this is not taken care of, your submission will not be graded.

You can make as many submissions you want if you want. [We might ask you to submit the script which was
used to generate the submission at any time].

In order to clear this project, you are required to clear both, Part 1 as well as Part 2 of this assignment.

Wish you all the best!

[Link] 2/2

Common questions

Powered by AI

Adhering to the specific guidelines like column names and row counts is crucial for the automatic grading system to function correctly. Deviations from these standards can lead to submission rejection or incorrect grading, impacting the evaluation process negatively .

The primary goal of the predictive model is to predict sales figures for counterfeit medicine operations using given data about these operations. This helps in understanding which operations are high net worth and thus should be targeted first to make enforcement more effective .

Deploying resources to counter counterfeit pharmaceutical rackets poses challenges such as ensuring resource allocation is not spread too thinly, which could render them ineffective. This challenge necessitates focusing on high-value illegal operations and using predictive models to strategically direct resources where they can have the most impact .

Project Process Guides are essential as they help participants understand the approach and methodology to construct and evaluate machine learning models. These guides offer structured directions and clarify the project's objectives, enhancing participants' ability to develop effective solutions .

Collecting data on counterfeit operations' characteristics allows for building predictive models that identify high-value illegal activities, which can then be targeted efficiently. This data-driven approach enables more precise enforcement actions, potentially reducing the spread and impact of counterfeit medicines .

Counterfeit medicines significantly harm global healthcare by being either fake, contaminated, or containing incorrect dosages of active ingredients. In developing countries, up to 30% of medicines are counterfeit, posing severe health risks to the population and straining healthcare systems. The World Health Organization (WHO) collaborates with Interpol to combat these issues, as these illegal medicines result in billions of dollars in criminal profits, yet the operations persist, challenging resource deployment .

The project proposes prioritizing efforts by focusing on illegal operations of high net worth instead of attempting to control all counterfeit operations. This strategic focus is based on data collection that will help predict sales figures associated with these illegal operations, allowing for more effective resource allocation .

Passing both parts of the project ensures that the participant has adequately understood and applied their knowledge to both theoretical and practical aspects of the task. This dual success criterion confirms readiness and capability to handle real-world predictive modeling challenges effectively .

Minimizing MAE is crucial because the project's scoring criterion is directly dependent on it. A lower MAE indicates a more accurate model, which is necessary for making reliable predictions about sales figures. This accuracy is pivotal to effectively directing efforts to combat high-value counterfeit operations .

The submission CSV must resemble 'sample_submission.csv' with identical column names and value types. The number of rows should match the test dataset exactly, ensuring no grading issues occur .

You might also like