The linear fit fails to fit the age profile because it does not really account for the changes in arrests rate which implies a large bias.
You need more than one comma, but a hyphen and some reorganisation might not go amiss.
Yours is like many things that people put on the site. It is a sentences which is infuriatingly correct in its grammar, and that because it is awkward and hard-work to read.
If it were me I would break it in three.
In the first nine words you have a repetition of 'fit'. Is that necessary? I would , at the very least, hyphenate the first one.
'The linear-fit fails to fit the age profile'. That is enough for a first sentence. It is short, snappy and makes its point; a single discrete thought.
'And that is because it doesn't adequately account for the changes in the arrests rate.' Note I have dropped 'really' in favour of 'adequately' as I find the former amorphous and vague. (Remember you are competing for the reader's attention with other things going on around them, office gossip in the background, or the TV, their kids, food cooking on the stove etc.) I have also added a definite article which I think helps.
Finally 'Hence it implies heavy bias'. 'Large' is not a word which leaps off the page and grabs the reader. I think 'heavy' stands a better chance.
So altogether 'The linear-fit fails to fit the age profile. And that is because it doesn't account for the changes in the arrest rate. Hence it implies heavy bias'
It is important to leave the conclusion you make The linear fit fails to fit the age profile as a sentence by itself if this is for a scientific manuscript.
The which question you ask can be avoided by splitting up the sentence into 3 parts as already suggested. Short sentences in technical writing allow readers to realize which parts of the data they are supposed to read along with the sentence. Any more than one piece of data per sentence makes digestion and comprehension difficult.