Lab4 CPIT 440 Data Mining and Warehouse
Lab4: Outlines Data Mining Process Data Gathering and Preparation (Preprocessing) –Techniques of the Data Preprocessing Data Integration Techniques Data Cleaning Techniques Data Transformation Techniques Data Discritization Techniques –Definition and Exercises CPIT 440 Data Mining and Warehouse
Data Mining Process CPIT 440 Data Mining and Warehouse
Data Gathering and Preparation The data understanding phase involves data collection and exploration. You can take a closer look at the data, you can determine how well it addresses the business problem. You might decide to remove some of the data or add additional data. Data preparation can significantly improve the information that can be discovered through data mining. CPIT 440 Data Mining and Warehouse
The data preparation phase covers all the tasks involved in creating the case table you will use to build the model. Tasks include data cleansing, binning and transformation. For example, –you might transform a DATE_OF_BIRTH column to AGE; –you might insert the average income in cases where the INCOME column is null. CPIT 440 Data Mining and Warehouse Data Gathering and Preparation
Data Preprocessing Techniques Data Integration Techniques: –Correlation (Numerical Data) by using Excel –Correlation (Categorical Data-Chi-Square Test) by using Excel Data Cleaning Techniques: –Fill the Missing Values by using ODM –Outlier Treatment for Reducing Noise by using ODM Data Transformation Technique: –Normalization by using ODM Data Discritization Technique: –Discritization by using ODM CPIT 440 Data Mining and Warehouse
Data Integration Technique Definition: Sometimes too much information can reduce the effectiveness of data mining. Data sets with many attributes may contain groups of attributes that are: Irrelevant attributes which is simply add noise to the data and affect model accuracy. –Noise increases the size of the model and the time and system resources needed for model building and scoring. CPIT 440 Data Mining and Warehouse
Data Integration Technique Or, correlated attributes that may actually be measuring the same underlying feature. –Their presence together in the build data can skew the logic of the algorithm and affect the accuracy of the model. To minimize the effects of noise, the technique like correlation is sometimes a desirable preprocessing step for data mining. CPIT 440 Data Mining and Warehouse
Data Integration Technique Exercises: Correlation (Numerical Data) by using Excel. Open Excel file Corr.xlsx Correlation Results will always be between -1 and 1 –1 = Positive Correlation –0 = No Correlation –-1 = Negative Correlation CPIT 440 Data Mining and Warehouse
Data Cleaning Technique CPIT 440 Data Mining and Warehouse 1.Fill the Missing Values by using ODM: –When building or applying a model, Oracle Data Mining automatically replaces missing values of –numerical attributes: with the mean, max/min, avg, specific value or zero values. –categorical attributes with the mode.
Exercise –Open ODM and import File demo_missing.csv Take a view on this file in the attribute length_of_residence there are some data missing; –Now we will apply a technique of data cleaning to fill out the missing data. From ODM open Data Transform Missing Value CPIT 440 Data Mining and Warehouse
Exercise –This will open Missing Value Transformation Wizard CPIT 440 Data Mining and Warehouse
Exercise –In the 4 th step of wizard Select the Column (attribute) on which you are going to apply missing Value technique and then press on Transform button. –You will see three option select Replace With – Mean. –Continue with next button till finish. CPIT 440 Data Mining and Warehouse
Exercise See the difference by using histogram, between Missing Data and after Fill Out Data. CPIT 440 Data Mining and Warehouse With Missing After solving Missing
Data Cleaning Technique 2. Outlier Treatment for Reducing Noise by using ODM: –A value is considered an outlier if it deviates significantly from most other values in the column. –The presence of outliers can have a skewing effect on the data and then can result in the inaccurate model –Outlier treatment methods such as trimming or clipping can be implemented to minimize the effect of outliers. CPIT 440 Data Mining and Warehouse
Exercise –Import File demo_outliers.csv Take a view on this file in the attribute years_details_listed, there are some outliers (Noise), means there are some values under this attribute which are very far from other. CPIT 440 Data Mining and Warehouse
Exercise –Now we will apply a technique of data cleaning to reduce this noise from the data. –Open Data Transform Outlier Treatment CPIT 440 Data Mining and Warehouse
Exercise – This will open Outlier Treatment Transformation Wizard –In the 4 th Step of wizard Select the Column (attribute) on which you are going to apply outlier treatment technique –then press std.deviation button then select edge/null values to be replaced with. CPIT 440 Data Mining and Warehouse
Exercise –Continue with next button till finish. CPIT 440 Data Mining and Warehouse
Exercise See the difference by using histogram, between Noisy data and after outlier treatment applied. CPIT 440 Data Mining and Warehouse
Data Transformation Technique: Normalization by using ODM: –Normalization is the technique that transforming numerical values into a specific range, such as [–1.0…1.0] or [0.0…1.0] CPIT 440 Data Mining and Warehouse
Exercise –Import File demo_original.csv Take a view on this file in the attribute family_income_indicator, we will apply normalize technique. CPIT 440 Data Mining and Warehouse
Exercise –Open Data Transform Normalize –This will open Normalize Transformation Wizard –In the 3 rd Step of wizard Select the Column (attribute) on which you are going to apply normalize technique and –then press Define button then select min-max transformation algorithm. CPIT 440 Data Mining and Warehouse
Exercise Continue with next button till finish. CPIT 440 Data Mining and Warehouse
Exercise Notice the difference by using histogram, before and after normalization. CPIT 440 Data Mining and Warehouse
Data Discritization Technique: Discritization by using ODM –Also called binning, is a technique for reducing the cardinality of continuous and discrete data. –It groups related values together in bins to reduce the number of distinct values. –Discritization can improve resource utilization and model build response time dramatically without significant loss in model quality. CPIT 440 Data Mining and Warehouse
Exercise –Import File demo_original.csv Take a view on this file in the attribute family_income_indicator, we will apply discritize technique. –Open Data Transform Discritize –This will open Discritize Transformation Wizard –In the 4 th Step of wizard Select the Column (attribute) on which you are going to apply discritize technique and –then press Equal Width button then write 10 number of bins. CPIT 440 Data Mining and Warehouse
Exercise Continue with next button till finish. CPIT 440 Data Mining and Warehouse
Exercise See the difference by using histogram, before and after discritization. CPIT 440 Data Mining and Warehouse Before After