Learning an Analytics tool

Number of data analytics tools such as SAS, R and SPSS are available including a few open source tools (details).

Framework to learn analytics tool

Framework to learn analytics tool

  1. Read
  2. Understand
  3. Manipulate
  4. Summarize
  5. Model or Analyze

1.      Read

The required data for an analysis may be available in various forms at various places. For example customer data may be available in Customer Relationship Management (CRM) systems in a database form such Teradata and DB2 tables but the offers details may be available in excel files.  A first and important step to proceed with data analysis is to bring data into an analytics tools. If Base SAS is being for data analysis, the data has to brought into SAS and create SAS datasets. Each of the analytics tools provide mechanism to connect to the input database or data file(s) and read the data.

2.      Understand

Once data is read and available in analytics tool format, great amount of time should be spend to understand the data. Understanding meanings of the variables/attributes and the values each attributes take is key. The variables could date such as birth date, categorical such as Marital Status, ordinal such as income level and continuous such as salary variables. Details on analytical variable types are available at..  Frequency and variable value distribution could be first steps to understand the data.

3.      Manipulate

For any meaningful analysis, data has to me manipulated and transformed before being used for the analysis, example creating age from date of birthday variable. To identify patterns and associations, one has to create derived variables and use an analytical or statistical technique. Additionally, combining various data tables such as customer demographic, product holding, credit history and transaction data in banking scenario.

4.      Summarize

Once the data is available in required form and level, they can be used for creating reports and dashboard for the senior executives or regulatory requirements. Also, the data can be used for developing predictive models or any other analysis and then be summarized.

5.      Model or Analyze

Aggregated data at the required level or/and combined have to ready for the comprehensive analysis and predictive modeling. If a predictive model to be build is to predictive a fraudulent transaction, the data have to be at that level and other information about the transactions such as timing and channel have to brought in. Whereas Model effectiveness models (Market Mix Models) may require data at a weekly level. But it may be helpful to get marketing promotion, campaign and competitor data for the model development.

Learn R yourself