Now, let's look at the details about the model, and see if it has used the same PROBLEM_TYPE and OBJECTIVE function which we mentioned while executed the CREATE MODEL commandĪccuracy of the Model and Prediction/Inference As the remaining steps of model training and compilation can take a longer time, it continues to run in the background.īut we can always check the status of the training using the STV_ML_MODEL_INFO function, and wait till the model_state becomes Model is Ready. Like for this problem, we are going to provide the PROBLEM_TYPE as multiclass_classification and OBJECTIVE as accuracy.Īs we have learnt in the PART-1 of the tutorial, the CREATE MODEL command operates in an asynchronous mode and it returns the response upon the export of training data to Amazon S3. When you provide this information while creating the model, Amazon SageMaker Autopilot chooses the PROBLEM_TYPE and OBJECTIVE specified by you, instead of tying everything. Now, being a data analyst, you may like to explicitly mention few of the parameters, like PROBLEM_TYPE and OBJECTIVE function. Similarly we can load the dataset for the testing( steel_fault_test.csv) in a separate table, steel_plates_fault_inference Once that is done, we can use COPY command to load the training data from Amazon S3 ( steel_fault_train.csv) to the Redshift cluster, in the table, steel_plates_fault.Īs always, we need to make sure that colum names of the table matches with the feature sets in the CSV training dataset file. We can open DataGrip(or whatever SQL Connector you are using) and create the schema and the table. So, the objective is to predict what is the fault the steel plate has ( Pastry, Z_Scratch, K_Scatch, Stains, Dirtiness, Bumps or Other_Faults)Īs we have seen in Part-1, since our dataset is located in Amazon S3, first we need to load the data in table. So, the problem in hand is a multi-class classification problem, where we need to predict the fault in the steel plate, given 7 different types of the faults that it can have. This dataset is related to the quality of steel plates, wherein there are 27 independent variables (input features) which comprises of various attributes of a steel plate and one dependent variable (class label) which can be of 1 of 7 types. You can download the dataset from this GitHub Repo. In this problem, we are going to use the Steel Plates Faults Data Set from UCI Machine Learning Repository. Now, we are going to see some of the advanced functionalities of Amazon Redshift ML which a Data Analyst or an expert Data Scientist can make use of, which offers more flexibility in terms of defining specific information, like which algorithm to use (such as XGBoost), specifying hyperparameter, preprocessor and so on. We also learnt about, how a database engineer/administrator could make use of Redshift ML to create, train and deploy a machine learning model using familiar SQL commands. Continuing our learning from where we left in the Part-1 of this tutorial series, where we discussed about Amazon Redshift briefly and dive deep into Amazon Redshift ML.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |