International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 07 Issue: 09 | Sep 2020 www.irjet.net p-ISSN: 2395-0072 © 2020, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 1738 Recognition of Human Activities using Machine Learning Algorithms Prof Jayashree M Kudari 1 , Abhilash Banerjee 2 1,2 Jain University ------------------------------------------------------------------------***----------------------------------------------------------------------- Abstract: Human activity recognition via wearable devices has been actively investigated during a huge variety of applications. Most of them, however, either target easy activities whereby whole-body movement is concerned or need a range of sensors to spot daily activities. During this study, we have a tendency to propose a personality's activity recognition system that collects information from associate off-the-peg smartwatch and uses a man-made neural network for classification. The planned system is additional increased victimization location data. We consider 6 activities, including both simple and dynamic activities. Keywords — Human activity recognition, machine learning I. INTRODUCTION Nowadays, smartphones became an essential tool for lifestyle, giving variety of options that transcend merely vocation or electronic communication capabilities. So as to let the user, perform totally different tasks, these mobile devices square measure equipped with several sensors that create them real sensing platforms able to extract relevant data. one in every of the foremost engaging eventualities during which such data are often exploited is act Recognition, wherever information captured by motion sensors, e.g., measuring device and rotating mechanism, are often analyzed to infer user’s current physical activity. Detecting and recording become conceivable with these moveable sensors that facilitate to acknowledge the topics frequently. Activities may also be place away once monitored in their favored surroundings [1]. II. METHODOLOGY A. Gathering Data The dataset was gathered and prepared by observing the activities of 30 volunteers within an age of 19-48 years [6]. Each person performing six different kind of activities are listed below: o WALKING o WALKING_UPSTAIRS o WALKING_DOWNSTAIRS o SITTING o STANDING o LAYING These data were gathered from UCI Machine learning library. The acquired dataset had been randomly dividing into two sets, where 70% of the volunteers were chosen for preparing information and 30% test information [7]. B. Feature Selection Feature selection is a vital construct in machine learning that is applied as a part of the pipeline. It’s the construct of mechanically or manually choosing a group of options that contribute to up the model and therefore the prediction output. This step is undertaken because it vastly impacts the performance of the model in terms of the build time moreover because the accuracy. Digressive options within the data set will negatively influence the training because it makes the model train on data that does not contribute to reaching the prophetic output. C. Data Visualization From this graph we can conclude that Nagpur Fig 1 shows us different human activities and their distribution. From this graph below we can see that sitting and standing are the highest amount of activities done by most of us. Fig 1. Different activity distribution Fig 2 shows us the difference between the static and dynamic activities.  Sitting, standing, laying can be considered as static activities were no motion is involved.