Data life cycle
we see in a brief but visual way the life cycle and its corresponding phases.
Data mining – data life cycle
crisp-dm implementation methodology
we now analyze this methodology in depth and everything it entails.
as with any project, it is important to know the objective that is being achieved in order to complete it.
To do this, we must establish the business objectives, evaluate the current situation, detect the objectives sought in data mining and the project milestones.
Data compression
this phase focuses on understanding and analyzing the capture of the data and its sources, understanding the information that describes them, and the relationships established between the different attributes.
Exploratory analysis will use graphs and calculations to determine the need for transformations of the captured data.
we will evaluate the quality
ensuring its suitability for correct modeling.
In understanding, we must verify the analytical organization objectives set in the previous phase.
Preparation
this phase requires the greatest consumption of time in the project.
After the two previous phases, we france phone number list will carry out the necessary tasks to select, clean, integrate and build our data sets for modeling.
Some activities in this phase are:
scaling of numerical variables to shift said variables to equivalent ranges.
Balancing records of the target variable in the case of classification algorithms.
Elimination of outliers or atypical values spam data from an observation that distort the model.
Removing or filling null or empty values.
If necessary, eliminated from input variables for presenting correlations with other variables or with those of low quality.
Grouping, binning, or other transformations of input variables.