#pragma GCC optimization ("O3")

#include <bits/stdc++.h>

using namespace std;

using ll = long long;
using vec = vector<ll>;
using mat = vector<vec>;
using pll = pair<ll,ll>;

#define INF (1LL<<61)
//#define MOD 1000000007LL 
#define MOD 998244353LL
#define EPS (1e-10)

#define PR(x) cout << (x) << endl
#define PS(x) cout << (x) << " "
#define REP(i,m,n) for(ll (i)=(m),(i_len)=(n);(i)<(i_len);++(i))
#define FORE(i,v) for(auto (i):v)
#define ALL(x) (x).begin(), (x).end()
#define SZ(x) ((ll)(x).size())
#define REV(x) reverse(ALL((x)))
#define ASC(x) sort(ALL((x)))
#define DESC(x) {ASC((x)); REV((x));}
#define BIT(s,i) (((s)>>(i))&1)
#define pb push_back
#define fi first
#define se second

template<class T> inline int chmin(T& a, T b) {if(a>b) {a=b; return 1;} return 0;}
template<class T> inline int chmax(T& a, T b) {if(a<b) {a=b; return 1;} return 0;}
class mint {
public:
    ll x;
    mint(ll x=0) : x((x%MOD+MOD)%MOD) {}
    mint operator-() const {return mint(-x);}
    mint& operator+=(const mint& a) {if((x+=a.x)>=MOD) x-=MOD; return *this;}
    mint& operator-=(const mint& a) {if((x+=MOD-a.x)>=MOD) x-=MOD; return *this;}
    mint& operator*=(const mint& a) {(x*=a.x)%=MOD; return *this;}
    mint operator+(const mint& a) const {mint b(*this); return b+=a;}
    mint operator-(const mint& a) const {mint b(*this); return b-=a;}
    mint operator*(const mint& a) const {mint b(*this); return b*=a;}
    mint pow(ll t) const {if(!t) return 1; mint a=pow(t>>1); return (t&1?*this*a:a)*a;}
    mint inv() const {return pow(MOD-2);}
    mint& operator/=(const mint& a) {return *this*=a.inv();}
    mint operator/(const mint& a) const {mint b(*this); return b/=a;}
};
istream &operator>>(istream& is, mint& a) {ll t; is>>t; a=t; return is;}
ostream &operator<<(ostream& os, const mint& a) {return os<<a.x;}

using mvec = vector<mint>;

vec convolution(vec a, vec b)
{
    struct LocalFunc {
        static ll modpow(ll a, ll n)
        {
            if(n == 0) return 1;
            ll t = modpow(a, n>>1);
            return (n&1?a*t%MOD:t)*t%MOD;
        }
        static void ntt(vec& a, bool rev=false)
        {
            ll i, j, k, l, p, q, r, s;
            ll size = a.size();
            if(size == 1) return;
            vec b(size);
            r = rev?(MOD-1-(MOD-1)/size):(MOD-1)/size;
            s = modpow(3, r);
            vec kp(size/2+1, 1);
            for(i=0; i<size/2; ++i) kp[i+1] = kp[i]*s%MOD;
            for(i=1, l=size/2; i<size; i<<=1, l>>=1){
                for(j=0, r=0; j<l; ++j, r+=i){
                    for(k=0, s=kp[i*j]; k<i; ++k){
                        p = a[k+r], q = a[k+r+size/2];
                        b[k+2*r] = (p+q<MOD)?(p+q):(p+q)-MOD;
                        b[k+2*r+i] = s*((p>=q)?(p-q):(MOD-q+p))%MOD;
                    }
                }
                swap(a, b);
            }
            if(rev){
                s = modpow(size, MOD-2);
                for(i = 0; i < size; i++) a[i] = a[i]*s%MOD;
            }
        }
    };
    
    ll size = a.size()+b.size()-1;
    ll t = 1;
    while(t < size) t <<= 1;
    vec A(t, 0), B(t, 0);
    for(ll i=0; i<a.size(); i++) A[i] = a[i];
    for(ll i=0; i<b.size(); i++) B[i] = b[i];
    LocalFunc::ntt(A);
    LocalFunc::ntt(B);
    for(ll i=0; i<t; i++) A[i] = A[i]*B[i]%MOD;
    LocalFunc::ntt(A, true);
    A.resize(max(a.size(), b.size()));
    return A;
}

vec ppow(vec A, ll n)
{
    if(n == 0) return vec{1};
    vec B = ppow(A, n>>1);
    return convolution((n&1?convolution(A, B):B), B);
}

int main()
{
    ll N;
    cin >> N;

    vec A(N-1);
    mint t(1);
    REP(i,0,N-1) {
        if(i) t *= i;
        A[i] = (mint(i+1)/t).x;
    }

    mint ans = t*mint(N).pow(N-2).inv();
    A = ppow(A, N);
    ans *= mint(A[N-2]);
    PR(ans.x);

    return 0;
}

/*



*/