No compromising on data quality

There’s a lot of things you can compromise on within data science.

Speed of data retrieval (slow server), the amount of data, the number of analysts.

But if there’s one that you should never compromise on, it’s data quality.

Why? Data quality is the backbone of any analysis.

Nothing brings more insecurity to the analysis than low data quality.

When you have high data quality, you can hire data analysts or increase server speed later.

But low data quality is insidious in every facet of the analysis.

• Can’t be reasoned with: When you can’t tell what the data represents, you can’t do a good analysis.
• Grinds everything to a halt: In some instances, data quality can increase by doing manual work. So-called data cleanup. But it is very time-consuming. So much so that the data may not be relevant anymore when it’s been “cleaned”.
• Leads you astray: Sometimes, the organization has to make do with what it has because it needs to make a decision. Low data quality may take you down the wrong path. Reversing an organizational decision is expensive, and may lead to employee fatigue.

Your data is only as valuable as its quality.

And lowering data quality even slightly leads to a dramatic decline in value.

– Samy