# Innovative Formula vs Traditional Formulas

--

Business Understanding is an essential matter and primary stage in data science lifecycle so that, all next stages form based on our Business understanding.

In shadow of technological development that world is witnessing, our business requires an innovate solutions.

Sometimes traditional formulas do not give us an accurate solution for current our issues and as simply that returns to timing of formulas created and nature of issues in that time.

let’s take a look for a real example I faced in my first contract.

In summer, Temperature degrees were raising in medical laboratories and that impacts on devices performance and exposes these devices for malfunctions.

We already had issues in A/C system and these issues were due to a designing process of A/C system. So, we could not solve them except by redesigns a new A/C system which can fit laboratories size and temperature emitted.

In that time, we were seeking to get Health Accreditation. It means we need a standard to showcase during assessment phase at least to intercede for us until we solved these issues.

As a team, we were thinking how we can create a quality standard to be an evidence during assessment phase.

Questions are the beginning of innovation way

We can summaries the problem statement:

How can we create a standard to be a showcase and an evidence used as a quality standard?

Firstly, we thought for calculating temperature degrees efficiency for each laboratory by using the tradition formula,

However, this formula did not be accurate.

Then, we categorized a quality standard to ** Good** and

**based on efficiency rate if rate greater than or equal 70% it is considered a Good otherwise Bad.**

*Bad*Let us suppose, the output value is the ideal temperature degree 22 Celsius and the input value is the measured temperature degree 25 Celsius, according to traditional formula, the efficiency is

It is an accepted rate. But what about if the measured temperature degree 29 Celsius, the efficiency is

This is not accurate it should be less than 70% as a bad rate. 29 Celsius considers as a high temperature degree according to most of our medical devices, most of them require room temperature degree less than or equal 25 Celsius.

After we tried a traditional formula and saw it did not give us an accurate result. We thought to create our own formula to get an accurate efficiency rate for temperature degree according to our categories [ Good, Bad].

Based on my knowledge in math, I took this challenge and tried many times until I reached the final form of efficiency formula with appropriate parameters.

Some parameters were setting with fixed value:

ideal temperature degree = 22 Celsius

actual temperature degree = measured temperature degree

ideal efficiency = 100%

actual efficiency = unknown

one temperature degree = 10% “ for each efficiency rate”

breakdown temperature degree = 32 Celsius

New Efficiency Formula:

actual eff. : actual efficiency

ideal eff. : ideal efficiency

actual temp. deg. : actual temperature degree

ideal temp. deg. : ideal temperature degree degree

one temp. deg. : one temperature degree

Let us suppose, actual temperature degree = 25 Celsius

It’s a logical and accurate rate according to temperature degree and considers a good rate.

But what about if the actual temperature degree = 29 Celsius

It’s an accurate rate according to temperature degree and considers a bad rate.

The next step after we have understood the problem and created efficiency formula was collecting data.

During you collected data, give more attention in this step.

## Why?

Based on the collected data you will take your decision.

Here you should ask yourself,

What features can they add into data?

Some features play critical role in the data and have high correlation with the target.

There were some features which we had added them to our data:

⁃ Laboratory name

⁃ Room number

⁃ Floor number

⁃ Devices number

⁃ Measuring temperature degree for each room

Here the screen shot of data after we collected “data frame”

Let’s take a look to these features and see which one has an impact on temperature degrees “ TemperuatureDegree ”.

How does it look like the correlation between Efficiency and Temperature degree?

The better way for describing features is visualizing for extracting results

**Results**

Efficiency Rate vs Temperature degree

- Strong inversely correlation between temperature degree and efficiency rate.
- If temperature degree exceeded 25 Celsius, the efficiency rate will become a bad otherwise be a good.

Temperature Degree and Efficiency Distribution

- [ 74% ] of Temperature Degrees were in the range from 25.7% to 29.2% as not accepted temperature degrees and [ 26% ] in the range from 22.2% to 24.1% as accepted temperature degrees.
- In contrast [ 74% ] of Efficiency Rates were in the range from 28% to 63% as low efficiency rates and [ 26% ]in the range from 70% to 98% as a high efficiency rates.

Number of devices per Room vs Temperature Degree

- Lower correlation between Devices Count and Temperature Degree which was due to some outliers.
- All laboratory rooms with devices more than four had high temperature degree “ bad rate ” and represented around [ 91% ] except one room had appropriate temperature degree “ good rate ”.
- In contrast, half of laboratory rooms with devices less than four had appropriate temperature degree “ good rate ” and other half had high temperature degree “ bad rate ”.

**Discussion**

There were two odds behind raising temperature degrees:

- First one refers that increasing number of devices in laboratories rooms might raise temperature degree as we saw when devices count were more than four.
- Second one maybe the area of rooms were wide and A\C system did not cover whole area. or the opposite were small and A\C system did not work effectively.

All of these odds are possible, but they need a right proof to ensure the correct decision-making.