For separate processes of market research, a large volume of data is stored at various levels. Depending on the level of the workflow and the data processing criteria, there are four primary forms of analysis descriptive, diagnostic, predictive and prescriptive. These four forms together react to all that a business wants to know from what’s going on in a business to what solutions it wants to implement to improve its functions.

Four forms of analytics are typically conducted in stages and no one type of analytics is claimed to be better than the other. They are interrelated, and each of them gives a particular perspective. With data being important to so many varied industries – from manufacturing to electricity grids – most businesses rely on one or more of these forms of analytics. With the best choice of computational tools, big data will provide businesses with richer insights.

Before we delve deeper into each of these, let’s describe four kinds of analytics:

**1) Descriptive Analytics:** explaining or summarising existing data using existing business intelligence software to help explain what is going on or what has happened.

**2) Diagnostic Analytics:** Rely on previous performance to assess what has happened and why. The product of the research is always the scientific dashboard.

**3) Predictive Analytics:** Stresses the estimation of a future result using mathematical equations and machine learning techniques.

**4) Prescriptive Analytics:** a form of predictive analytics that is used to prescribe one or more courses of action to evaluate the results.

Let ‘s explain this in a little more detail.

## 1. Descriptive research

This can be referred to as the easiest method of analytics. The scale of big data is beyond human understanding, and the first step thus requires the crunching of data into manageable chunks. The aim of this method of study is basically to summarise the results and understand what is going on.

Among other widely used terms, what people call advanced research or business intelligence is essentially the use of descriptive statistics (arithmetic operations, mean, median, max, percentile, etc.) on real results. It is said that 80 per cent of market analytics mostly includes explanations based on aggregations of past data. It is an important step to make raw data easier for customers, owners and managers to understand. This makes it possible to identify and overcome areas of strengths and limitations that can assist in the strategy.

The two key approaches involved are data aggregation and data mining, which claim that this approach is primarily used to explain the underlying behaviour and not to make any predictions. By mining historical data, corporations may evaluate customer preferences and commitments for their businesses that could assist for targeted ads, quality enhancement, etc. The methods used for this process are MS Excel, MATLAB, SPSS, STATA, etc.

## 2. Diagnostic Analytics

Diagnostic analytics are used to classify that something occurred in the past. It is distinguished by strategies such as drill-down, data discovery, data mining, and correlation. Diagnostic Analytics explores evidence more thoroughly in order to understand the underlying causes of incidents. It is useful to establish which causes and incidents led to the result. It mostly uses the odds, odds, and distribution of effects for interpretation.

Diagnostic analytics can help you explain whether revenues have declined or improved over a given year or so in a time series of sales results. However, this form of analytics has a limited capacity to offer actionable insights. It just gives an interpretation of causal connexions and sequences when thinking backwards.

Some diagnostic analytical approaches include significance of features, the study of key components, sensitivity analysis, and collaborative monitoring. This form of analytics also involves teaching algorithms for classification and regression.

## 3. Predictive Analytics

Predictive analytics are used, as described above, to forecast future results. Nevertheless, it is important to remember that whether an occurrence will occur in the future it can not predict; it only estimates the probabilities that the event will occur. In order to obtain the possible effects, a predictive model depends on the preliminary descriptive analysis.

Predictive analytics are important to build models such that current data can be interpreted to extrapolate or simply to forecast future data. One typical application of prediction analysis is to gather and evaluate all opinions posted on social media (existing text data) of order to forecast the feeling of the person in a given topic as positive , negative or neutral (future prediction).

The forecast of future data is based on current data, since it can not otherwise be accessed. If this model is accurately calibrated, nuanced revenue and marketing predictions may be supported. The regular BI takes things a step forward in forecasting accurately.

## 4. Prescriptive Analytics

Predictive analytics therefore require the creation and validation of predictive models. Predictive analytics depends on the use of algorithms for learning by computer such as random forests, SVMs, etc., Normally, businesses have to develop these models through qualified data scientists and machine-learning experts. Python, R, RapidMiner, etc. are the most common predictive analysis tools.

The forecast of future data is based on current data, since it can not otherwise be accessed. If this model is accurately calibrated, nuanced revenue and marketing predictions may be supported. The regular BI takes things a step forward in forecasting accurately.