I got asked this question in an interview and the interviewer only gave a hint. I can't find a convincing answer behind it.
Regarding the intuition part, what I partially know are the properties of additivity and $\ln(1+x)\approx x$
The interviewer subsequently asked me to calculate $\ln(105/100)$ without using a scientific calculator. The answer that was given to me was $5/102.5$. I checked that the answers match only approximately. As of now, it appears to be related to mean value theorem.
The $\ln$ function is continuous on $[100,105]$ and differentiable on $(100,105)$. Hence, by MVT, there exists a $c\in (100,105)$ such that
$\ln105-\ln100=\dfrac{1}{c}(105-100)=\dfrac{5}{c}$
Now, $100<c<105$. If we take an average of these two numbers, we can arrive at the approximation of 5/102.5.
This works on other examples $\ln\dfrac{107}{105}\approx \dfrac{2}{106.5}$
But why do we take an average ? Is this related to the shape of the plot of $\ln(x)$ or that $\ln(a)$ gives the area under the curve $1/x$ from $1$ to $a$
I feel that the correct answer should tie these two parts.
Please help.
The estimation made is the following:
$$\log(1+x) \approx x-\frac{x^2}{2}$$
which is compared to
$$\frac{x}{1+x/2} \approx x-\frac{x^2}{2}$$
hence
$$\log\left(\frac{105}{100}\right) \approx \frac{5}{102.5} \; .$$