Page 62 - CT_AI_Class-7
P. 62
DATA ANALYSIS
Data analysis is the process of examining data to understand what it means. It helps you make
better decisions by extracting useful information from data.
When you visualise data, it helps your brain quickly spot patterns and trends, making it easier to
understand the bigger picture. But data analysis goes a step further by using tools like computers
and mathematics to look deeper into the data for more detailed insights. After you visualise the
data, you can analyse it further to gain a better understanding.
Precision, Accuracy and Valid Data
When working with data, it’s important to understand precision and accuracy.
Precision refers to how detailed or exact a measurement is. Imagine an AI model that predicts
the price of a product. If the model gives you a price of `1000.50 every time you run it, the
predictions are precise because they are consistent. However, if the actual price is `1080.00,
then the model’s predictions are precise but not accurate.
Accuracy refers to how close your measurement is to the actual or true value. If an AI system
predicts the price of a product as `1800.05, which is very close to the true price of `1800.00, the
prediction is accurate because it’s nearly the same as the true value, even if it’s not exactly the
same.
Let’s use an example of measuring the length of a pencil, to understand precision, accuracy and
valid data:
Precision: You measure the pencil’s length using an AI-based sensor three times and each
time you get the measurement of 7.5 cm. The measurements are precise because they are
consistent. But if the actual length of the pencil is 8 cm, your measurement is precise but not
accurate.
Accuracy: If the pencil’s actual length is 8 cm and you measure it to be 7.9 cm with the AI
model, then the measurement is accurate because it’s close to the true value, even if it’s not
exactly the same.
Valid data is both accurate and precise. In the case of AI models, valid data would mean that
the model gives consistent and correct results. For your analysis and conclusions to be correct,
you need valid data.
fact bits
Loss of Precision in Digital Data In 1962, statistician John
Sometimes, when data is processed by computers, the precision Tukey defined data analysis
of the data can be lost. This is called loss of precision. Let us in his paper “The Future of
understand it from the example: Data Analysis.” His work
helped shape how we use
A number might originally be recorded as 25, but data analysis today.
after processing by an AI algorithm, it might appear as
25.000000.
60 Artificial Intelligence (CT & AI)-VII
