With the increasing demand for vehicles, predictive maintenance has gained more importance in the automobile industry. It has become however difficult to detect and recover the failure in-advance in vehicle due to the low scaled availability of sensors. However due to the advancements in the automobile industry, it has become very feasible to analyze and process the data i.e. the sensor data.Machine Learning techniques is used for prediction of failure. In this paper, we have presented an approach towards the fault prediction of the sub-systems of the vehicle. Sensorial data is collected while the bike is on the move. Now the data collected from the bike will be collected on both the conditions i.e. In normal condition when the bike is running efficiently and in the faulty condition, which is when a failure in a particular sub-system has occurred. The collected data is first sent to the android application via Bluetooth which further transmits the data onto the local server machine for the pre-processing and the analysis purpose. Various interesting patterns/outcomes are learned using the classifiers. These patterns are used to predict 2 possible outcomes i.e. 1.Root cause of failure 2.Time of failure. The root cause of failure can be predicted using the classification model and the possible time of failure can be predicted using the Regression model. These patterns can be later used to predict and detect the failures of other bikes which show identical behaviours. The ultimate goal of the approach is on increasing the vehicle’s up-time
