南航暑期国际课程大数据可视化第6讲综述.ppt

Name: AXXX AYYY Couse: CS-556 Instructor: Dr. Boris Kovalerchuk Machine Learning Field of study that gives computer the ability to learn without being explicitly programmed. We executed the program on different data-sets. For training the program we use training data, for testing the program we use test data and finally execute on real data. Supervised Machine Learning The learning where we provide the labeled data-set to train and test the program. Un-Supervised Machine Learning The learning where we do not provide the labeled data-set and the program/algorithm tries to find the correlation in the raw data. Implementation of Supervised Machine Learning We are using a data-set that contains the characteristics of cells sampled on women presenting (or not) a breast malignant tumor. The attributes are listed below. Clump Ucell Size Ucell Shape Mgadhesion Sepics Bnuclei Bchromatin Normnucl Mitosis Class Data visualization: We are using a unity tool (Lots of Lines) to visualize our data. This tool can draw the provided data using 4 visualization methods. Collocated Coordinates Radial (anchored) Paired Shifted Paired Parallel Coordinates Data-Visualization The below screenshot is taken after the data are loaded into the tool. Data-Visualization I have selected one particular visualization Collocated Pair and I have selected some vectors that have class equal to benign. Data-Visualization I have selected all the visualizations and I have selected some vectors that have class equal to benign and trying to see some pattern in the dataset. Data-Visualization To better visualize the data, lets load the data of two classes separately and try to see the similarities and differences. Below is the data visualization of class benign. Data-Visualization Below is the data visualization of class malignant. Data Visualization As we can see from the visualization of both classes separately that the visualization of both classes with parallel coordinates