Chetan Parekh

Ranch Hand

Posts: 3640

posted 12 years ago

Please consider the following situation

Download Time Before Optimization : 10 Seconds

Download Time After Optimization : 5 Seconds

There are two 2 logic

((10-5) / 10) * 100 = 50

You can say after optimization we have reduced 50% download time.

((10-5) / 5) * 100 = 100

If you compare Download Time After Optimization with the Download Time Before Optimization, you can say change is 100%.

I am with the seconds logic as it gives more optimized figure.

What is your opinion?

Is any calculation method wrong?

Download Time Before Optimization : 10 Seconds

Download Time After Optimization : 5 Seconds

There are two 2 logic

**First Logic:**((10-5) / 10) * 100 = 50

You can say after optimization we have reduced 50% download time.

**Second Logic:**((10-5) / 5) * 100 = 100

If you compare Download Time After Optimization with the Download Time Before Optimization, you can say change is 100%.

I am with the seconds logic as it gives more optimized figure.

What is your opinion?

Is any calculation method wrong?

My blood is tested +ve for Java.

Alan Wanwierd

Ranch Hand

Posts: 624

posted 12 years ago

There are 2 different things you are saying:

Firstly: After optimising the download is 50% faster

and secondly: Before optimising the download was 100% slower.

both statements are correct.

I'd be tempted to avoid confusion and say "performance much improved" !!

[ August 23, 2005: Message edited by: Adrian Wallace ]

Firstly: After optimising the download is 50% faster

and secondly: Before optimising the download was 100% slower.

both statements are correct.

I'd be tempted to avoid confusion and say "performance much improved" !!

[ August 23, 2005: Message edited by: Adrian Wallace ]

Chetan Parekh

Ranch Hand

Posts: 3640

Neeraj Dheer

Ranch Hand

Posts: 225

posted 12 years ago

You're saying that 5 is 1/2 of 10, and that 10 is twice 5. Same thing either way, it's just a question of how you want to spin it.

As i think Mark Twain said, "There are lies, damn lies, and statistics."

As i think Mark Twain said, "There are lies, damn lies, and statistics."

There are only two hard things in computer science: cache invalidation, naming things, and off-by-one errors

Ryan McGuire

Ranch Hand

Posts: 1136

8

posted 12 years ago

I disagree. When you talk about savings, it is generally understood that fractions/percentages are in terms of the original number. So cutting a 10 second time down to 5 seconds is a 50% improvement.

Let's say you cut the time down to 2 seconds. In terms of the original measurement, that's an 80% improvement. In terms of the new number, that would be a 400% improvement.

HOWEVER...

If you want the number to look better, you could use the

Files per minute before optimization: 6

Files per minute after optimization: 12

THAT would be 100% improvement. Just for comparison, let's say you got the transfer time for one file down to 2 seconds. That would be a rate of 30 files per minute. And that would be a speed improvement of 400%.

Some other tricks. If you're doing the optimization in phases, comparing the improvements made so far to the original number makes it look better.

So if your download time went from 10 sec to 5 sec to 2 sec. After the second round of optimizations, report that the speeds are now a 400% improvement over the original speed (instead of a 150% speed improvement over the first round of optimizations).

Another way to make the same measurement look good by measuring the "right" thing: Let's say that your new download software/hardware made the successful transfer percentage go from 99.85% to 99.95%. That's an improvement of only 0.10015%. Big deal. BUT if you measure

I love statistics.

Originally posted by fred rosenberger:

You're saying that 5 is 1/2 of 10, and that 10 is twice 5. Same thing either way, it's just a question of how you want to spin it.

As i think Mark Twain said, "There are lies, damn lies, and statistics."

I disagree. When you talk about savings, it is generally understood that fractions/percentages are in terms of the original number. So cutting a 10 second time down to 5 seconds is a 50% improvement.

Let's say you cut the time down to 2 seconds. In terms of the original measurement, that's an 80% improvement. In terms of the new number, that would be a 400% improvement.

HOWEVER...

If you want the number to look better, you could use the

*rate*of file transfer.

Files per minute before optimization: 6

Files per minute after optimization: 12

THAT would be 100% improvement. Just for comparison, let's say you got the transfer time for one file down to 2 seconds. That would be a rate of 30 files per minute. And that would be a speed improvement of 400%.

Some other tricks. If you're doing the optimization in phases, comparing the improvements made so far to the original number makes it look better.

So if your download time went from 10 sec to 5 sec to 2 sec. After the second round of optimizations, report that the speeds are now a 400% improvement over the original speed (instead of a 150% speed improvement over the first round of optimizations).

Another way to make the same measurement look good by measuring the "right" thing: Let's say that your new download software/hardware made the successful transfer percentage go from 99.85% to 99.95%. That's an improvement of only 0.10015%. Big deal. BUT if you measure

*failed*transfers instead, you've made those go from 0.15% to 0.05%, which is an improvement of 67%. Which would you rather report: an improvement (increase) of of .1% in successful transfers or an improvement (reduction) of 67% in failed transfer failures?

I love statistics.