Discovering hidden knowledge
Posted: Mon Dec 23, 2024 4:51 am
For modeling it is necessary to combine different techniques and technologies.
The choice of a particular programming language, a particular algorithm or fonts will determine the final result.
The knowledge acquisition process is based on choosing a categorical or numerical variable from our data set as the target variable.
The rest will be our input variables and from which we will try to determine the value of our target variable for future records of our sources.
Next, we will classify the algorithms and show some functionalities.
Supervised, unsupervised…
Supervised algorithms have data whose value of the target variable is known.
This allows us to train and test our model to make the necessary adjustments for its correct operation.
In this way, we can say that the algorithm learns from previous situations.
On the contrary, there are times when there are no data sets available whose target variable is properly labeled or quantified and we use the input data to classify, segment or group our records.
Currently, reinforcement learning algorithms have been defined where, in addition to the input algorithms, the current state is evaluated to calculate the action to be taken, seeking the greatest reward from the processing.
Algorithms according to the objective
Data mining can search for different objectives with clients , products, people or any other element of interest to be analyzed.
To do this, it will use different algorithms depending on the need, which may be supervised or unsupervised.
The range of options is extensive, as are the algorithms derived seeking to counteract some of their adverse effects.
A brief overview is provided below.
Regression problems
Sometimes we come across numerical target variables for which we want to predict or determine future values.
For them, we establish regression functions of the input variable with the target variable.
For example, we might want to predict a particular customer's future sales f dubai state name list or the year based on their attributes.
Data Mining - Regression Example
Clustering or segmentation
There are situations where it is necessary to segment or group a set of elements, such as customers, based on their attributes.
In this way, commercial strategies can be defined based on the segment in which a particular client is located.
Their attributes can be either personal (such as age) or based on their relationship with the organization (such as volume of purchases or type of products consumed).
Examples of this type of algorithms are Kmeans or hierarchical clustering .
The objective in this case is to obtain clusters where the elements of the same cluster have the minimum possible distance and the maximum distance from elements from which they are different.
The choice of a particular programming language, a particular algorithm or fonts will determine the final result.
The knowledge acquisition process is based on choosing a categorical or numerical variable from our data set as the target variable.
The rest will be our input variables and from which we will try to determine the value of our target variable for future records of our sources.
Next, we will classify the algorithms and show some functionalities.
Supervised, unsupervised…
Supervised algorithms have data whose value of the target variable is known.
This allows us to train and test our model to make the necessary adjustments for its correct operation.
In this way, we can say that the algorithm learns from previous situations.
On the contrary, there are times when there are no data sets available whose target variable is properly labeled or quantified and we use the input data to classify, segment or group our records.
Currently, reinforcement learning algorithms have been defined where, in addition to the input algorithms, the current state is evaluated to calculate the action to be taken, seeking the greatest reward from the processing.
Algorithms according to the objective
Data mining can search for different objectives with clients , products, people or any other element of interest to be analyzed.
To do this, it will use different algorithms depending on the need, which may be supervised or unsupervised.
The range of options is extensive, as are the algorithms derived seeking to counteract some of their adverse effects.
A brief overview is provided below.
Regression problems
Sometimes we come across numerical target variables for which we want to predict or determine future values.
For them, we establish regression functions of the input variable with the target variable.
For example, we might want to predict a particular customer's future sales f dubai state name list or the year based on their attributes.
Data Mining - Regression Example
Clustering or segmentation
There are situations where it is necessary to segment or group a set of elements, such as customers, based on their attributes.
In this way, commercial strategies can be defined based on the segment in which a particular client is located.
Their attributes can be either personal (such as age) or based on their relationship with the organization (such as volume of purchases or type of products consumed).
Examples of this type of algorithms are Kmeans or hierarchical clustering .
The objective in this case is to obtain clusters where the elements of the same cluster have the minimum possible distance and the maximum distance from elements from which they are different.