In this paper, we study a nonparametric estimation of random effects from the following distribution dependent SDE driven by fractional Brownian motions \(\begin{aligned} dX^j_t=\beta _jb(t,X^j_t,\mathcal {L}_{X_t^j})dt+\varepsilon \sigma (t, \mathcal {L}_{X^j_t})dB_t^{j,H}, ~X^j_0=x^j_0, ~0\le t\le T, \end{aligned}\) where \(j=1,2,\cdots,N\) , \(\mathcal {L}_{X^j_t}\) denotes the law of \(X^j_t\) , \(B_t^{1, H},\cdots,B_t^{N, H}\) are independent fractional Brownian motions with common Hurst parameter \(H\in (1/2,1)\) and \(\beta _j\) is random variable independent of \(B^{j,H}\) . This result extends the existing results of non parametric estimation to the case of distribution dependent. Moreover, we consider the estimation for the density function of \(\beta _j\) under weaker conditions of kernel function and study the corresponding asymptotic distribution.
