Learn practical skills, build real-world projects, and advance your career

Use of Ensemble Modelling to Reduce Fatalities in Aviation Industry

alt

A. INTRODUCTION

The data used in this project comes from a Kaggle competition, which attempts to reduce the fatalities in aviation industry by using
aircrew’s physiological data who may be distracted, sleepy or in other dangerous cognitive states.

Our challenge is to build a model to detect troubling events from aircrew’s physiological data. We'll use data acquired from actual pilots in test situations, and Our models should be able to run calculations in real time to monitor the cognitive states of pilots. With this model's help, pilots could then be alerted when they enter a troubling state, preventing accidents and saving lives.

B. DESCRIPTION OF THE DATA

All Data:

This Data consists of physiological data from eighteen pilots who were subjected to various distracting events.

The training set consist of 6 minutes set of controlled experiment collected in a non-flight environment. When members are subjected to these experiments, The sensors which are attached in different parts of their body transmits the real time data with us, Based on which we predict the behavioral condition of the member at that point of time under different experiences.

Here 4 sensors are used to prepare traing data--

  1. Electrocardiogram
  2. Respiration
  3. Galvanic Skin Response
  4. Electroencephalogram

Electrocardiogram:

• 3-point Electrocardiogram signal. The sensor had a resolution/bit of .012215 µV and a range of -100mV to +100mV. The data are provided in microvolts.

Respiration:

• A measure of the rise and fall of the chest. The sensor had a resolution/bit of .2384186 µV and a range of -2.0V to +2.0V. The data are provided in microvolts.

Galvanic Skin Response:

• A measure of electrodermal activity. The sensor had a resolution/bit of .2384186 µV and a range of -2.0V to +2.0V. The data are provided in microvolts.
"The galvanic skin response (GSR, which falls under the umbrella term of electrodermal activity, or EDA) refers to changes in sweat gland activity that are reflective of the intensity of our emotional state, otherwise known as emotional arousal."

Electroencephalogram :

•3-point Electrocardiogram signal. The sensor had a resolution/bit of .012215 µV and a range of -100mV to +100mV. The data are provided in microvolts.

Following four set of cognitive states of the member in experiments:

Channelized Attention (CA) --the state of being focused on one task to the exclusion of all others.

Diverted Attention (DA)-- is the state of having one’s attention diverted by actions or thought processes associated with a decision.

Startle/Surprise (SS)-- is induced by having the subjects watch movie clips with jump scares.

Baseline --- Normal state

Training Data:

  • 4.9 million samples in the training set (1.5Gb in size)
  • Experiments performed for each cognitive state to each individual members are performed.
  • Each experiment either gives cognitive state corresponding to that experiment or the Baseline as its state.

Test Data:

  • 19 Million samples in the test set (4.5 Gb in size)
  • Testing is done in simulated training vehicles.
  • Prediction of Cognbitive states based on that data is required.

C. Evaluation

Submissions are to done in CSV format. For each id in the test set, we have to predict a probability for each of the 4 possible cognitive states ( A = baseline / no event, B = SS, C = CA, D = DA).

Evaluation is done by Multi Class Log Loss between the predicted probabilities and the observed target.