- This topic has 5 replies, 2 voices, and was last updated 10 years ago by
.
- Posts
Hi john sir..hopes u r fine..plz help me with the follwing question..
The time taken by each of the first four batches is as follow:
Batch numbe Time taken
1 100
2 70
3 59
4. 55what is the learning rate closest to?
John sir ans is 84.3 in kaplan kit…bt i did’t get it ..john sir if 70 is an incrementl time then cum total time should be 170 and so cum avrg time per unit should b 85..so learning rate should b 85 ..bt when i add up 59 and 55 it comes to 114 which should b incrementl time to produce 3rd nd 4rth batch..in this way at cum output of 4 batches cum total time should b 284 and cum averg time per unit should b 71..bt at this level nw learning rate is diffrnt..71/85…..it comes to 83.53..this iz confusing me..it should not b like this..
If r is the learning that, then the average time per batch if we make 2, should be 100 x r
The average time if we make 4 should be 100 x r^2
The total time to make 4 is (100+70+59+55) = 284, and therefore the average time per batch is 284/4 = 71.
So 100 x r^2 = 71
r^2 = 0.71
r = 0.843 (or 84.3%)I suggest that you watch the free lecture on learning curves.
Thanx for your kind reply sir..john sir..i solved many question in the way which i explained above..This is the only questin which is confusing me……john sir. the point of confusion for me is that if we divide cum total time by cum output level of 4 ..it comes to 71 hours which is cum avrg time per unit..but when we divide cum total time by cum output level of 2..it comes to 85 hours which is cum avrg time per unit..in the question time for first unit is 100 hours..so in princple 85/100 comes to 85 percnt learing..same case should b with 71..but 71/85 does’t comes to 85…………..
jhon sir the way u explaind i knw due to your lectrs..bt the point of confusion is smthng else…plz guid meIn a perfect world, the learning rate would be constant and calculating it for the first 2 would give the same result as calculating it for the first 4.
However, in reality it would be a miracle if that happened, and so the more units we look at then the better the approximation would be. That is why it is better calculated over 4 units rather than over 2. (If it was available then calculating over 8 would be better still)
Again thnx for your kind reply and paitence..this time i got ur point..u r my ideal teacher..thnx?
- You must be logged in to reply to this topic.