Scribd Logo
Upload

Welcome to Scribd!

  • 268 views

    dp100 Cheat Sheet Machine Learning

    Uploaded by

    fares

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    dp100 Cheat Sheet Machine Learning

    Uploaded by

    fares
    268 views1 page

    Original Title

    dp100-cheat-sheet-machine-learning

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    268 views1 page

    dp100 Cheat Sheet Machine Learning

    Uploaded by

    fares

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 1

    Machine learning in a nutshell

    Machine learning uses algorithms to identify patterns within data. Patterns that are used to create a data model that can make predictions.
    The Azure for the Data Scientist course (DP-100) focuses on creating and using machine learning models with Azure Machine Learning.
    To understand the purpose of the exercises, a simplified overview is provided here.

    1. Define the problem


    Decide on what the model should predict and when it is successful.

    Classification Regression Time-series forecasting


    Predict a categorical value. Predict a numerical value. Predict future numerical values
    based on time-series data.

    Computer Vision Natural Language


    Classify images or Processing
    detect objects in images. Extract insights from text.

    2. Get the data


    Find data sources and get access. Azure Machine Learning connects seamlessly with the three most used Azure storages for data science.

    Azure Blob Storage Azure Data Lake Gen2 Azure SQL Database
    Object cloud storage. Unlimited object cloud storage. Relational cloud database.
    Uses flat namespace to store Uses hierarchical namespace for Used for tabular and
    unstructured data. granular access control. transactional data.

    3. Prepare the data


    Explore the data. Clean and transform the data based on the model’s requirements.

    Exploratory data analysis Feature engineering: Create validation set: Split


    (EDA): Analyse your data, get Transform the data to create the data into training and
    summary statistics, and features that will help the validation or test dataset to
    understand possible model to predict the target evaluate the model.
    correlations between variables. value.

    4. Train the model


    Choose algorithm and hyperparameters based on trial and error.

    Data
    Includes features (what influences the to be
    predicted value) and the target value (if it exists). Model
    Often stored as a binary file (e.g. pickle file).
    Algorithm Use on new data with same features to predict
    Based on the task (e.g. classification), the target value.
    different algorithms and hyperparameters
    can be tried.

    5. Integrate the model


    Use endpoint to generate predictions.

    Real-time predictions Batch predictions


    Create light-weight app to predict target Create pipeline to predict target
    value in real-time per new data value on new set of data
    measurement. measurements.

    6. Monitor the model


    Track the model’s performance.

    Data drift Evaluation metrics


    When new data differs significantly Keep track of the model’s performance.
    from training dataset. When the model’s predictions are
    increasingly incorrect.
    Retrain the model

    Learn more on:


    © Copyright Microsoft Corporation. All rights reserved.
    Create machine learning models Self-paced Microsoft Learn content on using Azure Machine Learning

    Explore visual tools for machine learning on Microsoft Learn Azure Machine Learning documentation

    You might also like