http://jira.mobinets.com/jira This file is an XML representation of an issue en-us 6.1.4 6159 19-11-2013 http://jira.mobinets.com/jira/browse/ACE-24 ACE <p>The objective is to use AI/Machine Learning to detect the relevant alarms from the irrelevant (transient) ones.</p> ACE-24

Relevant alarms detection system

Story Normal To Do Unresolved Unassigned Ahmed Osman Wed, 22 May 2024 08:28:36 +0300 Fri, 24 May 2024 12:42:54 +0300 Mon, 13 May 2024 00:00:00 +0300 0 1 <p>The first version "Relevant Alarms Detection v1" contains the data exploration and analysis of the alarms, and feature engineering.</p> <p>We will label an alarm as irrelevant if it is cleared within a short period of time, denoted as "n". The value of "n" should ideally be chosen by a domain expert. For the purpose of this study, we will use "n = 7 minutes".</p> <ul class="alternate" type="square"> <li><b>Modeling</b>:</li> <li>Split the data into 80% training and 20% testing sets.</li> <li>Trained a baseline Random Forest Classifier and evaluated its performance using precision, recall, f1-score, and accuracy metrics.</li> <li>Applied Stratified Shuffle Split to handle class imbalance and re-evaluated the model.</li> <li>Fine-tuned the model using Bayesian optimization to improve performance.</li> <li>Addressed class imbalance using SMOTE to oversample the minority class (Relevant alarms).</li> <li>Re-trained and evaluated the model post-oversampling, achieving significant performance improvements.</li> </ul> <ul class="alternate" type="square"> <li><b>Feature Importance Analysis</b>:</li> <li>Analyzed feature importance from the Random Forest model, highlighting the key contributors: Severity, Technical ID, FM Receive Time, and First Occurrence.</li> <li>Documented insights on the impact of each feature on model predictions.</li> </ul> <ul class="alternate" type="square"> <li><b>Results and Conclusions</b>:</li> <li>Achieved an F1-score, recall, precision, and accuracy of 96% with the oversampled model.</li> <li>Recommended future improvements, including the collection of more labeled data from domain experts to enhance model training.</li> </ul> Customer Dev. Actual Duration Gantt Options Last Comment <p style="display:inline"><b>Evaluation License Error:</b> EXPIRED</p> Original Estimate PDM Actual Duration Planned End Mon, 13 May 2024 00:00:00 +0300 Planned Start Mon, 13 May 2024 00:00:00 +0300 Rank 0|xtiau9: Rank (Obsolete) 9223372036854775807 VTDirections