Concepts of Data Analysis and Data Modelling

They are different but interlinked together

In this post, I would like to talk briefly about data analysis and data modelling because of two reasons. Firstly, I have written several posts concerning data analysis in a very random way, so it is necessary to talk about the main processes of the data analysis in systematic way. Secondly, I am planning to writing about some topics about modelling sooner.

Table of Contents

1. Data Analysis and Data Modelling

Data analysis and data modelling are two concepts, but usually interlinked together. In some sense, they mean more or less the similar thing.

(1) What is data analysis

In general, data analysis is:

a process of collecting, transforming, cleaning, and modelling data
to recover the useful information (Data mining)

(2) What is data modelling

In terms of modelling, it refers to

the process of analyzing and organizing the data elements
to find how the data elements relate to one another

2. Why Data Analysis and Data Modelling

The main reason includes, but not limits to, the following points:

Explore the useful patterns and information in the data
Understand developing direction and objectives
Discover the causes of certain events based on data findings
Understand problems facing
Present technical insights
Drive effective decision-making
Increase productivity

3. Examples of Data Analysis and Data Modelling

There are the 4 types of data analysis:

Descriptive Analysis: investigates the past data and finds what happened in the past.
Diagnostic Analysis: determines the reasons why that happened.
Predictive Analysis: predicts what would be likely to happen in the future (i.e. under past and current development trend).
Prescriptive Analysis: uses the information discovered from the previous 3 types of data analysis and provides a series of actions or strategies for different situations and scenarios (i.e. what-if question).

All these types of analysis typically rely on developing one or more models based on the data.

4. Process of Data Analysis and Modelling

The main processes include more or less the following aspects, of which many topics have been touched in the previous posts. You can reach them through the links in the text below.

Determine Objective: goals, reasons
Data Preparation: any processing of collecting, gathering, combining, structuring and organizing data for further analysis
Data Exploration: discover the useful information or/and problems through different operation, like missing values and outliers detection, data slicing, sorting, filtering, grouping, data visualization, etc.
Data preprocessing: clean data (such as missing values imputation, outliers treatment), integrate data, reduce data, transform data, encode dataset, find relations between data variables, normalizing data, split the data, etc.
Data Modelling: estimate or train model, validate it and test it, evaluate it, visualize the results, etc.
Model Deployment: apply it for prediction, strategy making under different scenarios, etc.
Result Communication: Report the results, publish the results, etc.