Aryma Labs

28/10/2019

Aryma Labs wishes you and your loved ones a very Happy and Prosperous Diwali!

30/08/2019

Data Science Tip:

In the previous post we highlighted the pitfalls of trusting Summary statistics blindly. There is another way users can be led astray and it is the scale of the charts.

Many a times novice analysts plot charts without paying attention to the scale of the chart.

A chart depicting growth of 20 % (from say 10,000 to 12,000) can be shown smaller relative to a chart showing growth of say 10 % (from 1000 1100).

Notice how on the left side, the growth looks flatter despite being 20% growth. But on the right side, the growth of 10% looks way steeper !!

Always remember to scale your charts properly. Wrong scale can lead to wrong insights.

For more tips, follow Aryma Labs on LinkedIn:
https://lnkd.in/fjxrFfw

25/08/2019

Data Science Tip:

Never trust summary statistics

06/08/2019

Data Science: Tip of the Day

Explaining the 'Linear' in Linear Regression

When people say Linear regression is linear, they mean that the model is linear in its parameters.
E.g. y = β0 + β1*x1 + β2*x2 has linear parameters (β1, β2) and is a linear equation.

If the equation is changed to y = β1*x1 + β2*(x1)^2, it will still be linear as the parameters are linear. Only the predictor is squared.

For more tips, follow Aryma Labs page:
https://lnkd.in/fjxrFfw

02/08/2019

Data Science: Tip of the day

What is Naïve about Naïve Bayes?

Naïve Bayes is called Naïve because it assumes that all features of the measurement are independent of each other.

For e.g. Suppose we were to classify an image into a dog and following are the features for measurement: length of the tail, shape of the ear, and body length.
Naives Bayes would assume that all the features are independent of each other which means there is no correlation between each feature. So, all the features contribute independently to the probability that the image is of a dog.

For more tips, follow Aryma Labs page:
https://lnkd.in/fjxrFfw

Also, visit our website: https://www.arymalabs.com/

29/07/2019

Data Science: Tip of the day

Do you know what is the difference between Principal Component Analysis and Factor Analysis?

PCA creates new set of variables from the old variables. The new variable /component is a linear combination of old variables with weights assigned to each of the old variable. Here the goal is to reduce the no. of variables to smaller components for ease of model building.

Factor Analysis on the other hand, assumes that the observed variables have a similar pattern of response and can be attributed to one latent factor. The latent factors formed explain the correlations between the variables. The goal is to test a model of latent factors causing observed variables.

For more tips, follow Aryma Labs page on LinkedIn:
https://lnkd.in/fjxrFfw



Also, visit our website:
https://www.arymalabs.com/

27/07/2019

Data Science: Tip of the day

Why Regression is called Regression?

The term was used by Francis Galton in his 1886 paper "Regression towards mediocrity in hereditary stature". In a way the statement refers to the property of "reversion to the mean" or "regressing to the mean". Hence the name Regression.

For more tips, follow Aryma Labs page on LinkedIn:
https://lnkd.in/fjxrFfw

Also visit our website:

https://www.arymalabs.com/

22/07/2019

Data Science: Tip of the day

Anova is a special case of Regression.

For more tips, follow Aryma Labs page!
https://lnkd.in/fjxrFfw

16/07/2019

Is your marketing leaving money on the table? Are you spending less on your marketing than you should? What if you could get the optimum ROI?

Marketing Mix Modeling can help you answer all these and more!!

Contact www.arymalabs.com/

The company Aryma Labs Pvt Ltd is named after two greats Aryabhatta (inventor of zero) and Madhava of sangamagrama (inventor of many calculus principles). The name is also clever pun of forecasting technique ‘Arima’.

14/07/2019

Data Science: Tip of the day

Understanding Null Hypothesis Significance Testing

In Null Hypothesis significance testing, one can either reject the Null Hypothesis or fail to reject it. Never can one "Accept it".

Many Aspiring Data scientists and even experts make this mistake.

The Null Hypothesis can never be accepted. The classic example is the legal system. The statement "Not guilty" does not mean the the defendant is "Innocent" . It merely means that there isn't sufficient evidence to prove 'beyond reasonable doubt' that the defendant committed the crime.

Similarly in statistical parlance, "Failing to reject the Null" does not mean "we can accept the Null".

For more tips, follow our page:

https://lnkd.in/fjxrFfw

Also, visit our website:
https://www.arymalabs.com/