Scribd Logo
Upload

Welcome to Scribd!

  • 6 views

    Mcon 1 2 ML Methodology MCON

    Uploaded by

    Baig

    Copyright:

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

    Mcon 1 2 ML Methodology MCON

    Uploaded by

    Baig
    6 views2 pages

    Original Title

    mcon_1_2_ML_methodology_MCON

    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
    6 views2 pages

    Mcon 1 2 ML Methodology MCON

    Uploaded by

    Baig

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    You are on page 1of 2

    Machine Learning Analysis for MCON

    MCON Team USA


    October 2024

    1 Introduction
    This report presents an analysis of predicting popular music tracks using var-
    ious machine learning models. The primary goal is to identify tracks, artists,
    and albums that are likely to become hits based on a variety of audio features
    such as danceability, energy, and loudness. The process integrates both data
    preprocessing and the application of predictive models to analyze the music
    dataset.

    2 Methodology
    The data preprocessing began by loading the music dataset, handling missing
    values, and scaling the features using StandardScaler to ensure uniformity. Pop-
    ularity was converted into a binary class for the classification task, where the
    goal was to predict whether a track would become a hit or not.
    We employed multiple machine learning models, including Logistic Regres-
    sion, Support Vector Machines (SVM), Decision Trees, and others. Logistic Re-
    gression, for instance, estimates the probability of a track being popular based
    on a linear combination of the input features. SVM, on the other hand, seeks to
    find the optimal hyperplane that separates hit tracks from non-hits. Decision
    Trees break the decision-making process into a series of conditional branches,
    making decisions based on features like danceability or tempo.
    After training the models, we computed probabilities or decision function
    scores for each track, predicting which were most likely to become popular. The
    top five tracks for each model were determined based on these scores, highlight-
    ing potential hits.
    In addition, statistical analysis was conducted by calculating the normalized
    means of ten different audio features. This provided a deeper understanding of
    the factors influencing popularity. The features included metrics such as energy,
    danceability, and tempo, which were examined to determine their correlation
    with popularity.

    1
    3 Results and Analysis
    The machine learning models successfully identified a number of tracks that
    were predicted to be hits. A detailed list of these tracks is available in the
    accompanying CSV file.
    Furthermore, the ranking of artists and albums was conducted by calculating
    the mean popularity and danceability scores, offering insights into which artists
    or albums had the most influence in the dataset. This ranking sheds light on
    key influencers within the music industry.
    The normalized mean scores for the ten selected audio features demonstrated
    significant variance across tracks. For instance, tracks with higher energy and
    danceability were often predicted to be more popular. This suggests that certain
    characteristics may be more indicative of a track’s potential success.

    4 Conclusion
    Through the application of various machine learning models, this analysis pro-
    vided valuable insights into the factors that contribute to a track’s popularity.
    By combining the predictions of multiple models and conducting detailed sta-
    tistical analyses of audio features, we were able to effectively predict top tracks
    and rank influential artists and albums. This comprehensive approach offers a
    robust framework for predicting music trends in the future.

    You might also like