Data Science Updates

Imagine we are making a self-driving car and we are trying to localize its position in an environment. The sensors of the vehicle can detect cars, pedestrians, and cyclists. Knowing the location of these objects can help the car make judgements, preventing collisions. But other than knowing the location of the objects, the car needs to predict their future locations so that it can plan what to do ahead of time. For example, if it were to detect a child running towards the road, it should expect the child not to stop. The Kalman filter can help with this problem, as it is used to assist in tracking and estimation of the state of a system.

The car has sensors that determine the position of objects, as well as a model that predicts their future positions. In the real world, predictive models and sensors aren’t perfect. There is always some uncertainty. For example, the weather can affect the incoming sensory data, so the car can’t completely trust the information. With Kalman filters, we can mitigate the uncertainty by combining the information we do have with a distribution that we feel more confident.

Usage of the Kalman filter acknowledges that some error and noise is implicit within both the prediction and measurement of the variables we’re interested in tracking.

The filter does not assume all errors are Gaussian, but as cited from the Wikipedia description, the “filter yields the exact conditional probability estimate in the special case that all errors are Gaussian.” Recall, that the Gaussian is characterized by two parameters: the mean (mu) and the width of Gaussian, namely the variance (sigma squared).

Imagine we are making a self-driving car and we are trying to localize its position in an environment. The sensors of the vehicle can detect cars, pedestrians, and cyclists. Knowing the location of these objects can help the car make judgements, preventing collisions. But other than knowing the location of the objects, the car needs to predict their future locations so that it can plan what to do ahead of time. For example, if it were to detect a child running towards the road, it should expect the child not to stop. The Kalman filter can help with this problem, as it is used to assist in tracking and estimation of the state of a system.

The car has sensors that determine the position of objects, as well as a model that predicts their future positions. In the real world, predictive models and sensors aren’t perfect. There is always some uncertainty. For example, the weather can affect the incoming sensory data, so the car can’t completely trust the information. With Kalman filters, we can mitigate the uncertainty by combining the information we do have with a distribution that we feel more confident.

Usage of the Kalman filter acknowledges that some error and noise is implicit within both the prediction and measurement of the variables we’re interested in tracking.

The filter does not assume all errors are Gaussian, but as cited from the Wikipedia description, the “filter yields the exact conditional probability estimate in the special case that all errors are Gaussian.” Recall, that the Gaussian is characterized by two parameters: the mean (mu) and the width of Gaussian, namely the variance (sigma squared).

Hi,

First of all thanks for the question. Data Science as a subject has multiple layers. A great way to get started would be to brush up basic statistical concepts. Fundamental concepts of probability, distributions and Hypothesis Testing can get you started working on certain Data Science  problems. For example, if a marketing campaign has taken place during Diwali and some customers are given some offers , what is the best way to measure the impact of the campaign? How can we say that the impact was statistically significant ? 

Problems of supervised and unsupervised learning however requires deeper understanding. Multiple online available sources can help you get clarity on concepts. The key thing is to know what topics to search for. Here is a starting  list of topics you would want to research on with regards to unsupervised learning

   Principal Component and Factor         Analysis

   Cluster Analysis

    Conjoint Analysis

    Multi Dimensional Scaling

Topics on Supervised Learning could be and not limited to 

   Regression (OLS and GLM) 

   Non Parametric methods such as

      Decision Trees

      Random Forest

       SVM

       Boosting Techniques

Once you get the gist of the concepts on these topics , it will be much easier for you to grasp the discussions in any training program if you join one. In terms of programming Python is agreat platform .A few years ago, SAS was almost always important but things have changed now. SkLearn offers great set of packages in Python to play with. I'd encourage you to try and apply algorithmic approaches to create your own functions in Python. In conclusion, in my experience Data Science has been a journey and not a cluster of skills. Over time it will become a part of your thought process and will guide you every step of the way. Keep learning and keep practicing. 

If you need to discuss anything further - reach out to me.

Aritra

Data Science is a vast field. First of all you should learn statistics which is very important in Data Science field. Then you need to learn about basic Data Analytics and concepts. Languauges like SAS, R , Python can be chosen and learnt if you have the right concepts. All of these are based on the similar Data Science algorithms and learning a language is not difficult if you have your concepts very strong and domain knowledge. If you want to learn Data Science in a structured manner, please contact me as I have a very comprehensive course which can also be modified to suit individual needs and requirements.