Quality Assurance and Project Management

Mar 8 2015   3:28PM GMT

Six Steps Of An Analytics Project

Jaideep Khanduja Jaideep Khanduja Profile: Jaideep Khanduja

Tags:
Analytics
Business Analytics
data model

We discussed, in the end, about Analytics project in the previous post, from where we will take it further. If we look at the key steps of an Analytics project as below, we can very well correlate it with the project lifecycle where more or less the steps resemble with the various steps we following in a software project. Here are the various steps to be followed in an analytics project:

1. Problem Statement or Definition:

Understanding business problem well is the most important factor so as to ensure that you move in the right direction. Definitely like other projects, analytics projects also have tight release dates. Business problem will tell you what exactly is the issue that you are going to resolve in alignment with the business context. Your immediate task after understanding the business problem will be to gather the relevant data belonging to considerably large period. At times, when you are lucky, your customer will give you relevant data along with the business problem statement. But if not, then you will have to hunt for the data. Your customer’s business problem will need to be translated into analytics problem. By defining the analytics problem, you will be clearly able to state what you are going to predict at the end of this analytics project after building analytics model. A clarity or rather a customer sign off on this will be very helpful in later closings.

2. Data Crunching or Exploration:

Data familiarization and exploration are the next important steps to follow.

3. Information Building:

Once you have understood the data well, you need to go for some data modelling. This will be an exhaustive exercise in which you will identify the missing links and apply data modelling techniques. Your must be quite familiar with certain terms like values, outliers, variables, regression, decision tree, clustering etc. to master this phase of the project.

4. Data Modelling:

Once your data is in place after you have done some homework in your data crunching and data building, the next step is data modelling. In this phase, you follow a technique run-observe in a repeated manner. Here you will have multiple iterations of running a model, evaluating the results, perform some calibration and re-run, re-observe till you reach a satisfaction level of results desired.

5. Validation or Confirmation

The final set of shortlisted models (not more than three usually) now have to pass the validation process. Here you will put in a fresh set of data in the data model built to ensure that the model is not specific to previous data or conditions only and holds good to produce a similar kind of resultant.

6. Deployment and Monitoring:

Out of the final set of shortlisted models, the best one is chosen for deployment and monitoring the results. Monitoring of results is important to check the model behavior in real life scenario.

 Comment on this Post

 
There was an error processing your information. Please try again later.
Thanks. We'll let you know when a new response is added.
Send me notifications when other members comment.

Forgot Password

No problem! Submit your e-mail address below. We'll send you an e-mail containing your password.

Your password has been sent to:

Share this item with your network: