h*******o
发帖数: 1114 | 1 有人用过Eigenvector的PLS toolbox吗？有问题想请教一下 |
l*********1
发帖数: 351 | 2 这个要收费的吧.还有很多其他的pls toolbox的.好像有个丹麦产的不错. |
j******0
发帖数: 258 | 3 I used it a lot. You can post your question here. |
h*******o
发帖数: 1114 | 4 非常感谢！就是说我有40多个Raman谱图（是从零时间到全部生成产物的终点），是水
解反应A+H2O---〉B+C+D——〉C+D，B是intermediate,最后全部生成C（完全水解产物
），没有calibration spectra,在这种情况下，用PCA建Model还需要做cross
validation吗？如何分析data以求得rate constant(假定反应为first order)?
【在 j******0 的大作中提到】
: I used it a lot. You can post your question here.
|
j******0
发帖数: 258 | 5 You start from A and end with C and D. Then you know how many PCs are you
are looking in the spectral data. In such case, you don't need cross
validation to find the optimal number of PCs which is also the number of
compounds in the system.
Your spectrum is a function of reaction time. You can try evolving factor
analysis to locate the time point where new reaction product starts to
emerge. You can also get the concentration profile of every reactant. This
will help you get the reaction rate. |
h*******o
发帖数: 1114 | 6 非常感谢！我们老师说因为包括起始物中间体和最终两个产物，应该有4 PCs，而且应
该能get principle components for each of the materials and the loadings for
each as a function of time.我就在想怎样去做这个loadings for each as a
function of time，是指用PLS-toolbox可以生成4个新的谱图，每个谱图有40多个
overlaid spectra(我的original spectra 有40多个?就好像是#1谱图代表起始物从0时
刻到最终时刻的谱图？但是不对阿，横坐标是时间啊，loadings在这里指什么？不是指
pure component spectra吗？那如何与时间关联？用那种3-D的图解释吗？那我应该生
成4种3-D的图对吗？横坐标是wavenumber,纵坐标还是intensity,然后Z轴排列不同的图？
谢谢！
【在 j******0 的大作中提到】
: You start from A and end with C and D. Then you know how many PCs are you
: are looking in the spectral data. In such case, you don't need cross
: validation to find the optimal number of PCs which is also the number of
: compounds in the system.
: Your spectrum is a function of reaction time. You can try evolving factor
: analysis to locate the time point where new reaction product starts to
: emerge. You can also get the concentration profile of every reactant. This
: will help you get the reaction rate.
|
