2026年9月1日火曜日

t検定 (t-test)

t検定は,2つのグループの平均値に「統計的に意味のある差(有意差)」があるかどうかを判定するための統計手法.得られたデータに平均値の差があっても,それが「たまたま (偶然) 」 生じたものかを見極める.

具体的には,2郡の分散が大きく異ならない場合,平均・分散→統合 (プールした) 分散→t値→p値を求め,例えば有意水準5%の場合,p値 (有意確率) が基準の0.05より大きければ,「意味のある差がない」 と判断する.

Example Question (Oxford AI SL P408 Example7)

Mr Arthur gives his two chemistry groups the same test. He wants to find out whether there is any difference between the achievement levels of the two groups.

The results are:

a)   Write down the null and alternative hypotheses, 

   a) の解答は下の"Oxford TextのSolution"の通り

b)   Find the t-value and p-value for a t-test at the 5% significance level, 

c)   Write down the conclusion to the lest.






GDCで求めた値は以下の通り.
t分布からP(T24>0.235)
p=2P(T24>0.235)

Excelの 「データ分析」 の結果は以下の通り.

Excelでt値をいきなり求めるには 「データ分析」 を用いる方法しかない.

Excelでp値をいきなり求めるには,T.TEST(配列1,配列2,2,2)とすればp=0.816229568が得られる.

Oxford Textbook の Solution

a)   $H_0$ : $\overline{x_1}= \overline{x_2}$ (there is no differencebetween the grades in Group 1 and the grades in Group 2)

      $H_1$ : $\overline{x_1} \neq \overline{x_2}$ (there is a difference between the grades in Group 1 and the grades in Group 2)

b)   t-value =-0.235,   p-value = 0.816

c)   0.816 > 0.05, so you accept the null hypothesis: there is no significant difference between the two groups.

Notice that the two groups do not need to be the same size. This will be a two-tailed test as you want to know whether Group 1 is better or worse than Group 2.


****


 (Oxford AI HL P603 Q12) 

Lisa measures the arm spans (the length from finger tips on one hand to finger tips on the other hand, when arms are held horizontally outstretched) of students in two mathematics classes. She is interested to see whether the mean arm span of class A is the same as the mean arm span of class B. Lisa's data is shown in the table below.


In this question, you will use a t-test to compare the means of the two groups at the 10% level of significance. You may assume the data is normally distributed and the standard deviations are equal between the two groups, (等分散を仮定してよいと指示している)

a)  (i)   State the null hypothesis.
     (ii)  State the alternative hypothesis,

   a) の解答は下の"Oxford TextのSolution"の通り

b)  Find the associated p-value for this test,
c)  State, giving a reason, whether Lisa should accept the null hypothesis.




GDCで求めた値は以下の通り.
t分布からP(T20>1.273)
p=2P(T20>1.273)

Excelの 「データ分析」 の結果は以下の通り.

Excelでp値をいきなり求めるには,T.TEST(配列A,配列B,2,2)とすればp=0.217565256が得られる.

Oxford Textbook の Solution
a)   (i)   $H_0$ : $\overline{x_A}= \overline{x_B}$     
      (ii)   $H_0$ : $\overline{x_A} \neq \overline{x_B}$
b)   p = 0.209 (3 sf)  (不等分散を仮定した解答になっている)
c)   As 0.209 > 0.1 there is no evidence to reject the null hypothesis at the 10% level of significance.

2026年7月27日月曜日

カイ二乗検定 chi-squared test ($\chi^2$ test)

アンケートや実験などにおいて、2つのデータ群の関連性を判定する検定.例えば,2つのデータ群が独立である (関連がない) という仮説を立てておき,そうなる確率 (p値) が極めて低いときは 独立でない (関連がある) と判定して仮説を棄却する.

具体的には,データ→期待度数→カイ二乗検定統計量(カイ二乗値)→p値を求め,有意水準5%の場合,p値が0.05より大きければ独立である (関連がない) と判定する.

◆ Oxford  AI  SL  P396  Q2 より

A survey was conducted to find out which type of bread males and females prefer. Eighty people were interviewed outside a baker's shop and the results are shown below. 

Using the x2 test at the 5% significance level, determine whether the favourite type of bread is independent of gender. 

a) State the null hypothesis and the alternative hypothesis.

$H_0$ : 帰無仮説 性別と好きなパンの種類は独立である(関連がない)
$H_1$ : 対立仮説 性別と好きなパンの種類は独立でない(関連がある)

b) Show that the expected frequency for female and white bread is approximately 15.9. 

女性で白い小麦粉のパンが好きな人の期待度数は $$\frac{41}{80}\times\frac{31}{80}\times 80=15.8875≒15.9$$

c) Write down the number of degrees of freedom. 

$(4-1)(2-1)=3×1=3$

d) Write down the $\chi ^2$ test statistic and the p-value for this data. The critical value is 7.815. 
critical value (臨界値) とは,「偶然では起きにくいと判断する境界線の値」 のことで,この場合は自由度3のときの値.

例えば,女性で白い小麦粉のパンが好きな人の観測度数(O)と期待度数(E)のズレは,
$$\frac{(O-E)^2}{E}=\frac{(17-15.8875)^2}{15.8875}=0.07790...$$
すべてのズレを合計したカイ二乗検定統計量 (カイ二乗値)は,
$$\chi^2=\sum\frac{(O-E)^2}{E}=2.1186...≒2.12$$
$f(x)=\chi^2$pdf$(x,3)$

p-value$=\int_{2.1186}^{\infty} f(x) dx=\chi^2$cdf$(2.1186,1$E$99,3)=0.5481...≒0.548$

Excelでいきなりp値を求めるには,CHISQ.TEST(観測値の配列,期待度数の配列)を使えばp=0.548165119 が得られる.

e) Comment on your result.

この場合,2.12<7.815   i.e.   0.548>0.05  なので,帰無仮説は棄却されず,性別と好きなパンの種類は独立である(関連がない)


◆ Oxford   AI   SL   P602   Paper 1   Q10 より

As part of a study into colour preference, Alicia surveyed 100 students at her school. She recorded each student's gender, and their favourite colour from blue, red, brown or black. Alicia's results are shown in the table below.

Alicia conducted a $\chi ^2$ test for independence at a 5% level of significance.

a)  State the null hypothesis.

 $H_0$ : Favourite colour is independent of gender.

b)  Write down the associated p-value.



p-value$=0.02764...≒0.0276$

Excelでいきなりp値を求めるには,CHISQ.TEST(観測値の配列,期待度数の配列)を使えばp=0.0276434が得られる.

c)  State, giving a reason, whether the null hypothesis should be accepted.

0.0276<0.05 なので,reject $H_0$ i.e. 好みの色と性別は独立でない(関連がある)

2026年7月2日木曜日

積分因子

小説「本と鍵の季節」より

1階線形常微分方程式の積分因子 $M(x)$ が $\exp \left( \int{P(x)}dx \right)=e^{\int{P(x)}dx}$ である理由 $$\frac{dy}{dx}+P(x)y=Q(x)\tag{1}$$ (1)の両辺に $M(x)$ を掛けると,$$M(x)\frac{dy}{dx}+M(x)P(x)y=M(x)Q(x)\tag{2}$$もし(2)を次の形にすることができれば以下のように微分方程式が解ける.$$M(x)\frac{dy}{dx}+M'(x)y=M(x)Q(x)\tag{3}$$$$\frac{d}{dx}(M(x)y)=M(x)Q(x)$$$$M(x)y=\int M(x)Q(x)dx$$$$y=\frac{\int M(x)Q(x)dx}{M(x)}\tag{微分方程式(1)の解}$$そのために以上のことができる積分因子 $M(x)$ を求めよう.(2)と(3)を比較して,

$$M'(x)=M(x)P(x)$$

$$\frac{dM}{dx}=M(x)P(x)$$

$$\int\frac{1}{M(x)}dM=\int P(x)dx$$

$$\ln{|M(x)|}=\int P(x)dx$$$$M(x)=\exp \left( \int P(x)dx \right)\tag{積分因子}$$

[簡単な例] 次の1階線形微分方程式を解く

$$\frac{dy}{dx}+y=e^x$$

積分因子 $\exp \left( \int{1}dx \right)=\exp(x)=e^x$ を両辺に掛けると,

$$e^x\frac{dy}{dx}+e^xy=e^{2x}$$

$$\frac{d}{dx}(e^xy)=e^{2x}$$

$$e^xy=\int e^{2x}dx=\frac{1}{2}e^{2x}+C$$

$$y=\frac{1}{2}e^{x}+Ce^{-x}$$

2026年6月25日木曜日

IB Math Textbook

Oxford AA HL の訂正

P835

d ii)    $v_n+_1$ではなく,$v_{n+1}$

d iv)   $u_n=2^n-1$ではなく,$u_n=(u_1+\frac{b}{a-1})a^{n-1}-\frac{b}{a-1}$

1 e)    解答がない.$a=2, b=1, u_1=1$ を上式に代入すると,$u_n=2^n-1$ となることが確認できる.

P807

Exercise  8H 1 e)  $y=e^{-\frac{x^2}{2}} \left(\frac{1}{2} e^{x^2}+C\right)$ではなく,$y=e^{-\frac{x^2}{2}} \left(\frac{1}{3} e^{\frac{3}{2}x^2}+C\right)$ または $=\frac{1}{3} e^{x^2}+Ce^{-\frac{x^2}{2}}$

Oxford AI SL の訂正

P396   

Ex8C   Q2   b)   17.9 ではなく 15.9

P670

Paper 1   1 B)    V_0 の式は1500000ではなく,1000000なので,正解は175550.90

                        V_10=175550.90*(0.9)^10=61210.81


Hodder Education AA HL の訂正

p474

11 b ii) -3,  16,  -65ではなく-113

P475

1 b  i) グラフは単調増加

P476

c vi) k<0 ではなく,k≦0

d) 0.1286ではなく,0.1283

2026年6月16日火曜日

GDC

PolyRoots,  cPolyRootsは,TI-Nspire CX に はあるが,TI-84 Plus CE にはない.TI-84 Plus CE では Numeric Solver (Equation  solver) を使うが実数解のみ.

Analize Graph => zero は複数あってもいちいちboundを決めて1つづつ求める.

複素数zのn乗根をすべて求めるときは,まず(1/n)乗して偏角θを求め,その後 2θ/n, 4θ/n, ......としていく (1周が2πなので).

TI-Nspire CX IIで定義域(範囲)を指定してグラフを描くには,数式を入力した後,[ctrl] + [=] を押して「 | 」を選んで入れてから定義域を入力する.

Listに数列を入れるには,stat>edit>L1の最上で,list >ops>seq を使う

2026年6月8日月曜日

三項間漸化式

 $a_1=1$, $a_2=2$のとき,次の三項間漸化式を解く.$$a_n=a_{n-1}+a_{n-2}$$初期値を $a_0=1$, $a_1=1$として次の漸化式を解いても同様の結果を得られるので,計算簡素化のため,こちらを解くことにする.$$a_{n+1}=a_n+a_{n-1}\tag{0}$$ある実数$\alpha$と$\beta$を用いて(0)が次のように変形できれば,さらに下の(5)(6)のように変形できて漸化式が解けるので,$$a_{n+1}-\alpha a_n=\beta(a_n-\alpha a_{n-1})\tag{1}$$$$a_{n+1}-\beta a_n=\alpha(a_n-\beta a_{n-1})\tag{2}$$(1)と(2)をどちらも次式に変形し,$$a_{n+1}-(\alpha+\beta) a_n+\alpha\beta a_{n-1}=0\tag{3}$$(3)が(0)に一致することから係数比較すると,$\alpha+\beta=1$,$\alpha\beta=-1$ となるので,$\alpha$と$\beta$は次式 (特性方程式) の解であることがわかる.$$t^2-(\alpha+\beta)t+\alpha\beta=0\tag{4}$$$\alpha+\beta=1$ より $1-\alpha=\beta$,$1-\beta=\alpha$となるので,(1)と(2)は次のように変形できる.

$a_{n+1}-\alpha a_n=\beta(a_n-\alpha a_{n-1})=\dots=\beta^n(a_1-\alpha a_0)=\beta^n(1-\alpha)=\beta^{n+1}\tag{5}$

$a_{n+1}-\beta a_n=\alpha(a_n-\beta a_{n-1})=\dots=\alpha^n(a_1-\beta a_0)=\alpha^n(1-\beta)=\alpha^{n+1}\tag{6}$

(6)から(5)を引くと,$$(\alpha-\beta) a_n=\alpha^{n+1}-\beta^{n+1}$$$$a_n=\frac{1}{\alpha-\beta}(\alpha^{n+1}-\beta^{n+1})$$(4)の解を求めて$\alpha$と$\beta$の値を代入すれば求める第n項を得る.$$a_n=\frac{1}{\sqrt{5}}\left\{ \left(\frac{1+\sqrt{5}}{2}\right)^{n+1}-\left(\frac{1-\sqrt{5}}{2}\right)^{n+1} \right\}$$フィボナッチ数列の一般的な初期値 $a_0=0$, $a_1=1$なら第n項は次式になる.$$a_n=\frac{1}{\sqrt{5}}\left\{ \left(\frac{1+\sqrt{5}}{2}\right)^{n}-\left(\frac{1-\sqrt{5}}{2}\right)^{n} \right\}\tag{ビネの公式}$$





2026年3月22日日曜日

百五減算のしくみ

百五減算とは,年齢$N$を3, 5, 7でそれぞれ割った余り$a, b, c$ から言い当てる方法で,具体的には,$70a+21b+15c$ から繰り返し105 (3, 5, 7のLCM) を引き,105未満になったらそれを$N$とする.

<しくみ>

以下,$a\equiv b \pmod n$は, 「$a-b$ は $n$ の倍数である」 ことを意味する.例えば,$17\equiv 2 \equiv {-1}\pmod 3$

$7\times5=35$,$7\times3=21$,$5\times3=15$なので,$N\equiv35a+21b+15c \pmod {105}$と予測できるが,$$35a+21b+15c\equiv 2a \pmod 3$$ $$35a+21b+15c\equiv b \pmod 5$$ $$35a+21b+15c\equiv c \pmod 7$$であり,ここで$a$は0, 1, 2のいずれかなので,$2a$は0, 2, 4のいずれかになり,これを3で割った余りは0, 2, 1のいずれかになるから,0以外は$a$と一致しない.従って,$$〇a+21b+15c\equiv a \pmod 3$$となるようにしたい.そのためには$a$の係数は35の倍数かつ$\equiv 1 \pmod 3$にしなければならないので,$$70a+21b+15c \pmod {105}$$とすれば良い.

例えば,$N$が$(a,b,c)=(1,2,3)$のとき,$$70\cdot1+21\cdot2+15\cdot3=157 \equiv 52 \pmod {105}$$言い換えると,3で割った余りが1で,5で割った余りが2で,7で割った余りが3となる105未満の数は52であることがわかる.

<7×15=105減算>

年齢$N$を7, 15でそれぞれ割った余り$a, b$から言い当てることもできる.

$N\equiv15a+7b \pmod {105}$と予測できるが,$$15a+7b\equiv a \pmod 7$$ $$15a+7b\equiv 7b \pmod {15}$$であり,ここで$b$は0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14のいずれかなので,$7b$は0, 7, 14, 21, 28, 35, 42, 49,  56, 63, 70, 77, 84, 91, 98のいずれかになり,これを15で割った余りは0, 7, 14, 6, 13, 5, 12, 4, 11, 3, 10, 2, 9, 1, 8のいずれかになるから,0, 5, 10以外は$b$と一致しない.従って,$$15a+〇b\equiv b \pmod {15}$$となるようにしたい.そのためには$b$の係数は7の倍数かつ$\equiv 1 \pmod {15}$にしなければならないので,$$15a+91b \pmod {105}$$とすれば良い.

例えば,$N$が$(a,b)=(3,7)$のとき,$$15\cdot3+91\cdot7=682 \equiv 52 \pmod {105}$$言い換えると,7で割った余りが3で,15で割った余りが7となる105未満の数は52であることがわかる.

<5×21=105>でも同様にできるし,積が105でなくてもできる.さらには,互いに素な2数や3数でもできるが,積が小さい数だと年齢当てには不向き.


2026年3月16日月曜日

2次正方行列の積の可換条件

$A=\left(\begin{matrix}a&b\\c&d\\\end{matrix}\right)$,$B=\left(\begin{matrix}e&f\\g&h\\\end{matrix}\right)$とする.

$$AB=\left(\begin{matrix}a&b\\c&d\\\end{matrix}\right)\left(\begin{matrix}e&f\\g&h\\\end{matrix}\right)=\left(\begin{matrix}ae+bg&af+bh\\ce+dg&cf+dh\\\end{matrix}\right)$$ $$BA=\left(\begin{matrix}e&f\\g&h\\\end{matrix}\right)\left(\begin{matrix}a&b\\c&d\\\end{matrix}\right)=\left(\begin{matrix}ae+cf&be+df\\ag+ch&bg+dh\\\end{matrix}\right)$$

成分比較して,

$bg=cf$ …①  $af+bh=be+df$ …②  $ce+dg=ag+ch$ …③

①より $\frac{b}{f}=\frac{c}{g} ⇔ b:c=f:g$ …④ 

②より $(a-d)f=(e-h)b ⇔ \frac{a-d}{e-h}=\frac{b}{f}$ …⑤

③より $(a-d)g=(e-h)c ⇔ \frac{a-d}{e-h}=\frac{c}{g}$ …⑥

⑤⑥より $\frac{a-d}{e-h}=\frac{b}{f}=\frac{c}{g} ⇔ \left(a-d\right):b:c=\left(e-h\right):f:g$

これは④も満たすので,可換になるための必要十分条件は $$\left(a-d\right):b:c=\left(e-h\right):f:g$$


2026年3月15日日曜日

正接定理でなく正弦定理で求める方法

小説 「écriture 新人作家・杉浦李奈の推論」より

正接定理でなく正弦定理で求める方法

$\triangle OMS$において正弦定理より
$$\frac{r}{\sin(2r-\frac{x}{2})}=\frac{1}{\sin(2r+\frac{x}{2})}$$
$$r\sin\left(2r+\frac{x}{2}\right)=\sin\left(2r-\frac{x}{2}\right)$$
$$r\left(\sin2r\cos\frac{x}{2}+\cos2r\sin\frac{x}{2}\right)=\sin2r\cos\frac{x}{2}-\cos2r\sin\frac{x}{2}$$
$$(1+r)\cos2r\sin\frac{x}{2}=(1-r)\sin2r\cos\frac{x}{2}$$
$$\cos2r\sin\frac{x}{2}=\frac{1-r}{1+r}\sin2r\cos\frac{x}{2}$$
$$\frac{\sin\frac{x}{2}}{\cos\frac{x}{2}}=\frac{1-r}{1+r}\cdot\frac{\sin2r}{\cos2r}$$
$$\tan\frac{x}{2}=\frac{1-r}{1+r}\tan2r$$
$$\frac{x}{2}=\arctan \left( \frac{1-r}{1+r} \tan 2r \right)$$
$$\stackrel{\huge\frown}{BC}=x=2\arctan \left( \frac{1-r}{1+r} \tan 2r \right)$$

2026年1月5日月曜日

最小二乗法 回帰直線導出

最小二乗法とは,データ分布を最もよく表す関数 (回帰曲線) を求める方法で,具体的には実際のデータ $\{(x_i, y_i) | i=1, \cdots ,n \}$ と予測関数の値 $\{y_i=f(x_i) | i=1, \cdots ,n \}$ の差の二乗の総和 (残差平方和 Residual Sum of Squares)$$RSS=\displaystyle\sum_{i=1}^{n} \{y_i-(f(x_i )\}^2$$が最小になるような関数 $y=f(x)$ を求める方法です.

最も簡単な例として,回帰曲線が一次関数 $y=ax+b$ になる場合の$a$と$b$を求めてみます.この場合は,$$RSS=\displaystyle\sum_{i=1}^{n} \{y_i-(ax_i +b)\}^2$$の最小を考えることになります.

偏微分を使うのが簡単なのですが,高校までの知識なら平方完成を使います.他のサイトで紹介されている計算は複雑なものが多いのですが,残差の二乗の平均$$\frac{RSS}{n}=\frac{1}{n}\displaystyle\sum_{i=1}^{n} \{y_i-(ax_i +b)\}^2$$を計算することで,よりシンプルに求めることができます.ここで,$\overline{x}$ は $x$ の平均を表します.

\begin{eqnarray}
\frac{RSS}{n}&=&\frac{1}{n} \displaystyle\sum_{i=1}^{n} (y_i^2+a^2x_i ^2+b^2-2ax_iy_i+2abx_i-2by_i )\\&=&\overline{y^2}+a^2\overline{x^2}+b^2-2a\overline{xy}+2ab\overline{x}-2b\overline{y}\tag{1}\\&=&b^2+2(a\overline{x}-\overline{y})b+a^2\overline{x^2}-2a\overline{xy}+\overline{y^2}\\&=&\{b+(a\overline{x}-\overline{y})\}^2-(a\overline{x}-\overline{y})^2+a^2\overline{x^2}-2a\overline{xy}+\overline{y^2}\\&=&\{b+(a\overline{x}-\overline{y})\}^2-a^2\overline{x}^2+2a\overline{x}\cdot\overline{y}-\overline{y}^2+a^2\overline{x^2}-2a\overline{xy}+\overline{y^2}\\&=&\{b+(a\overline{x}-\overline{y})\}^2+\left(\overline{x^2}-\overline{x}^2\right)a^2+2(\overline{x}\cdot\overline{y}-\overline{xy})a+\overline{y^2}-\overline{y}^2
\end{eqnarray}

ここで,$\overline{x^2}-\overline{x}^2=s_x^2$ ($x$の分散),$\overline{xy}-\overline{x}\cdot\overline{y}=s_{xy}$ ($x$と$y$の共分散) なので,

\begin{eqnarray}
\frac{RSS}{n}&=&\{b+(a\overline{x}-\overline{y})\}^2+s_x^2a^2-2s_{xy}a+s_y^2\\&=&\{b+(a\overline{x}-\overline{y})\}^2+s_x^2\left(a^2-2\frac{s_{xy}}{s_x^2}a\right)+s_y^2\\&=&\{b+(a\overline{x}-\overline{y})\}^2+s_x^2\left(a-\frac{s_{xy}}{s_x^2}\right)^2-\frac{s_{xy}^2}{s_x^2}+s_y^2
\end{eqnarray}

以上より,$a$と$b$が次の値のときに$\frac{RSS}{n}$の値が最小になる,すなわち$RSS$の値が最小になるということが分かりました.$$a=\frac{s_{xy}}{s_x^2}, b=\overline{y}-a\overline{x}$$ 因みに,$x$と$y$の相関係数は $r=\frac{s_{xy}}{s_x s_y}$ なので,$a=\frac{s_y}{s_x}r$ とも表すことができます.(2026年現在の高校数学Bの教科書ではこちらの方が書かれています)

高校では未習ですが,式$(1)$を$a$,$b$で偏微分して$=0$とおいた2式を連立させれば,容易に上の解を得ることができます.

                                             $a$で偏微分  $2a\overline{x^2}-2\overline{xy}+2b\overline{x}=0$
                                             $b$で偏微分  $2b+2a\overline{x}-2\overline{y}=0\ $      

<簡単な例>

データ $A (2, 3),B (4, 7),C (9, 11)$

$ \overline{x} = 5,\overline{y} = 7,s_x^2 = \frac{26}{3},s_{xy}=\frac{ 28}{3},a = \frac{s_{xy}}{s_x^2}= \frac{14}{13},b = \overline{y} - a \overline{x} = \frac{21}{13}$

回帰直線は $ y = \frac{14}{13}x + \frac{21}{13}$

2025年12月14日日曜日

ロジスティック方程式

次の微分方程式 (k:比例定数,0<y<1) をロジスティック方程式といいます.$$\frac{dy}{dx}=ky(1-y)$$変数分離してこれを解くと$$\int\frac{1}{y(1-y)}dy=k \int dx$$

$$\int\left(\frac{1}{y}+\frac{1}{1-y}\right)dy=kx +C$$$$\ln y-\ln(1-y)=kx +C$$$$\frac{y}{1-y}=e^{kx+C}$$$$y=(1-y)e^{kx+C}$$$$y\left(e^{kx+C}+1\right)=e^{kx+C}$$$$y=\frac{e^{kx+C}}{e^{kx+C}+1}$$$$y=\frac{1}{1+e^{-(kx+C)}}$$$x$軸方向の拡大縮小を1倍,平行移動を0にすれば,ロジスティック関数の標準形になります.$$y=\frac{1}{1+e^{-x}}$$

これを$x$軸方向に$k$倍,$y$軸方向$L$倍拡大縮小し,$x$軸方向に$p$,$y$軸方向$q$平行移動した式は次式になり,多くの自然現象,社会現象のモデル化に応用されています.$$y-q=\frac{L}{1+e^{-k(x-p)}}$$

[参考]

Logistic function
https://en.wikipedia.org/wiki/Logistic_function