Descriptive statistical processing
From raw data to information
The colletion process alows to obtain raw data. Data collection depends on the collection quality.
Soyculto system uses the concept of multiple option question , directly inspired from the Lickert scale. With 4 options per question, the scale is characterized as “forced choice“. With this “closed” scale collection system, the data has the required quality level to be processed and the information retrieval is made.
Besides, there are many data in the organization (either internal or external). In any case, the raw data has little value. Nevertheless, in processing, organizing or categorizing the data may reveal value added information.
The investigation methodologies, the statistics processing and the results may differ according to the customers and their objectives. However, the structure of the raw data collected by the Soyculto system, allows an optimum descriptive processing.
The descriptive statistics processing takes into account many elements such as:
- the average,
- the standard deviation,
- the variance,
- the distribution…
These results provide a description of the studied group (population) in various dimension at a specific time. The “photography” evaluates the market, the customers or the targetted audience in the dimensions required by the objective.
Datamining and predictive processing
If the descriptive statistics provides a “snapshot” at a specific time, the advanced statistics offers a scalable evaluation.
The advanced or inferential statistics, generates models, inferences and predictions evaluating the data and taking into account the random and uncertainty factors of the observations.
The inferential statistics processing makes :
- hypothesis testing,
- association descriptions (correlation),
- Regression analysis…
The obtained results offer an evolutionary description of the studied group (the population). The “sequence” evaluates the evolution of the market, the customers or the targeted audience in the aspects required by the objective.
Datamining and knowledge discovery
The datamining consists in extracting useful and unknown information, implicitly available, through a process of survey and exploration of the data.
Technics such as decision trees, statistics models and clustering are used to discover new knowledge useful for:
- refining a segmentation,
- analyzing (buying or consumption) habits ,
- identifying upselling or crosselling ideas,
- determining a behavior (in internet, of usage…),
- elaborating new strategies…
The datamining is particularly effective with Soyculto multi dimensional system. The topics are totally customizable. This allows to evaluate all the necessary dimensions. As the data are correctly formatted from the origin, the datamining analysis and processing are more optimized.